Background

Acute kidney injury (AKI) is a common complication after non-cardiac surgery with an incidence ranging from 6.8 to 39.3% according to different patient populations [1, 2]. Several underlying susceptibilities, procedures, or exposures have been identified to be risk factors of postoperative AKI occurrence, such as older age, chronic kidney disease, diabetes, sepsis, major surgery, and hemodynamic instability [3]. Recent evidence demonstrated that AKI was independently associated with longer length of hospital stay and higher rate of 30-day hospital readmission, 1-year end-stage renal disease, and mortality with more severe stage of AKI relating to poorer outcomes after non-cardiac surgery [4, 5]. Unfortunately, an effective treatment for AKI in the intensive care unit has not been established [6], strongly suggesting that early recognition of and adjusting for risk factors would be beneficial for high-risk patients.

Hypoalbuminemia is a well-established risk factor for increased morbidity and mortality in acutely ill patients [7]. The association between hypoalbuminemia and AKI is consistently evident in many observational studies conducted across different clinical settings, mainly focusing on infectious diseases, cancer, cardiac surgery, and transplant surgery [8,9,10,11,12]. Although the underlying mechanisms for this association are not fully elucidated, serum albumin may play a protection role in the maintenance of renal perfusion, preservation of proximal tubular integrity and function, binding of endogenous toxins and nephrotoxic drugs, prevention of oxidative damage, and delivery of protective lysophosphatidic acid [13, 14]. Therefore, we assumed that preoperative hypoalbuminemia might be associated with an increased risk of AKI following non-cardiac surgery. However, limited data are currently available on this topic [15,16,17,18]. Thus, this study aimed to investigate the association between preoperative serum albumin concentration and AKI occurrence in high-risk patients following non-cardiac surgery.

Methods

Ethics and consent

Ethical approval (2018–137) was provided by the Clinical Research Ethics Committee of Peking University First Hospital on July 4, 2018. Because of the retrospective nature of the study and no patient follow-up was performed, the ethics committee agreed to waive written informed consent. This study was performed in accordance with Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) criteria [see Additional file 1: Table S1].

Patients

The study period was from July 1, 2017, to June 30, 2018, with the following inclusion criteria: (1) adult patients (age ≥ 18 years), (2) undergoing non-cardiac surgery, (3) admitted to the surgical intensive care unit (SICU), and (4) at a high risk of postoperative AKI. High-risk patients referred to as patients having at least one of the following conditions: (1) preoperative comorbidities, including hypertension, diabetes mellitus, coronary heart disease, congestive heart failure, cerebrovascular disease, chronic kidney disease, lung disease, or liver disease; (2) major surgery, defined as surgery duration ≥2 h; (3) ongoing organ dysfunction, defined as the sequential organ failure assessment (SOFA) score ≥ 2 from one single organ system. Patients with any of the following criteria were excluded: (1) chronic kidney disease stage 5 or requiring long-term dialysis; (2) surgery involving kidney, such as nephrectomy, partial nephrectomy, nephroureterectomy, or kidney transplantation; (3) AKI events before surgery; and (4) incomplete clinical data. Our medical center is a teaching hospital affiliated with a university, which provides tertiary care and has about 1600 beds.

Definitions of outcomes

The primary end point was postoperative AKI development. Postoperative AKI and its severity was defined according to Kidney Disease Improving Global Outcomes criteria using the maximal change in serum creatinine compared with the preoperative baseline values and urine output during the first 7 postoperative days [19].

The secondary end points were the postoperative use of mechanical ventilation (MV) and its duration, length of ICU and postoperative hospital stay, number of postoperative complications other than AKI, total cost, and in-hospital mortality.

Other main postoperative complications were pulmonary infection, pleural effusion, pulmonary atelectasis, respiratory failure, surgical bleeding, new-onset arrhythmia, acute myocardial infarction, congestive heart failure, hemodynamic insufficiency, stroke, acute liver injury, disseminated intravascular coagulation, ileus, anastomotic leakage, intra-abdominal abscess, wound infection, wound dehiscence, urinary tract infection, sepsis, digestive tract bleeding, and venous thromboembolism [see Additional file 2: Table S2].

Other data collection

Patients’ data were searched through the electronic medical records system of our hospital. Perioperative data were collected including demographic characteristics (age, sex), body mass index (BMI), medical history, American Society of Anesthesiology (ASA) physical status classification, as well as preoperative nephrotoxin exposure. Other data collected were preoperative clinical laboratory data, such as hemoglobin, albumin, baseline serum creatinine, and B-type natriuretic peptide (BNP). Intraoperative data included type and duration of surgery, emergency surgery, duration of anesthesia, maximal lactate, minimal hemoglobin, estimated blood loss, use of vasopressors, volume of artificial colloids infusion, and fluid balance. Postoperative data before AKI included new onset of nephrotoxin exposure, sepsis, use of vasopressors, minimal hemoglobin, maximal lactate and BNP, perioperative blood transfusion, and non-renal SOFA score within 24 h of ICU admission.

Statistical analysis

Preoperative serum albumin concentration were firstly compared between patients with and without AKI by independent samples t-test. Then to determine its cutoff value for postoperative AKI occurrence, receiver operating characteristic curve analysis was performed. The patients were divided into two groups according to the occurrence of AKI or hypoalbuminemia. Quantitative variables with normal distribution were compared by independent samples t-test; numeric data with abnormal distribution were compared by Mann-Whitney U test. Qualitative variables were compared by chi-squared test or Fisher’s exact test. Time-to-event data were analyzed with Kaplan-Meier survival analysis, with difference between groups compared by log- rank test. After testing for collinearity, perioperative variables with P < 0.10 and number of events > 10 in the univariate analyses for AKI occurrence were included in multivariate logistic regression model (backward stepwise method) to identify independent risk factors for AKI. Furthermore, baseline variables unbalanced between patients with or without hypoalbuminemia were entered into propensity score matching. Then, patients were matched 1:1 based on their scores using nearest-neighbor matching with the tolerance being 0.02. Thereafter, logistic regression analysis was performed to find out the association between hypoalbuminemia and AKI. Two-sided P values < 0.05 were regarded as statistically significant. The SPSS v21.0 software package was used for statistical processing. (SPSS Inc., Chicago, IL, USA).

Results

During the study period, a total of 971 patients at a high risk of AKI undergoing non-cardiac surgery were admitted to SICU; among them, 729 met the inclusion/exclusion criteria and were included in the final statistical analysis (Fig. 1). Baseline and perioperative data were listed in Tables 1 and 2. Among the enrolled patients, 188 (25.8%) developed postoperative AKI stages 1, 2, and 3, which accounted for 21.9, 1.6, and 2.2%, respectively. Patients with postoperative AKI had significantly lower level of preoperative albumin compared with patients without AKI (36.7 ± 6.3 vs 39.3 ± 6.0 g/L, P < 0.001). The cutoff value of preoperative serum albumin for postoperative AKI occurrence was 37.5 g/L determined by the Youden index (P < 0.001, area under the curve [AUC] = 0.624) with a sensitivity of 0.54, specificity of 0.67, and positive predictive value of 0.36 [see Additional file 3: Figure S1].

Fig. 1
figure 1

Flow diagram of the study

Full size image
Table 1 Preoperative variables
Full size table
Table 2 Intra- and postoperative variables
Full size table

The incidence of AKI in patients with serum albumin < 37.5 g/L (35.9% [98/273]) was significantly higher than those with serum albumin ≥37.5 g/L (19.7% [90/456]) (P < 0.001). Univariate analysis showed that preoperative serum albumin < 37.5 g/L was strongly associated with the occurrence of postoperative AKI (OR 2.277; 95% CI 1.624–3.194; P < 0.001). In the multivariate logistic regression model (backward), preoperative serum albumin < 37.5 g/L was identified to be independently associated with postoperative AKI (OR 1.892; 95% CI 1.238–2.891; P < 0.001). Other independent risk factors for AKI included age (OR 1.018; 95% CI 1.004–1.033; P = 0.013), radiocontrast exposure (OR 1.843; 95% CI 1.031–3.293; P = 0.039), baseline creatinine (OR 1.016; 95% CI 1.008–1.025; P < 0.001), ASA classification (OR 1.719; 95% CI 1.193–2.477; P = 0.004), and intraoperative use of vasopressors (OR 1.680; 95% CI 1.065–2.648; P = 0.026) (Table 3). After matching for age, BMI, history of chronic kidney disease, preoperative hemoglobin, and ASA classification, 161 pairs of patients with or without hypoalbuminemia were well balanced in their baseline variables except for malignant neoplasm and ASA classification [see Additional file 4: Table S3; Additional file 5: Table S4]. Logistic regression analysis once again revealed that preoperative serum albumin < 37.5 g/L was independently associated with postoperative AKI (OR 3.085; 95% CI 1.649–5.771; P < 0.001) [see Additional file 6: Table S5].

Table 3 Independent risk factors for postoperative AKI
Full size table

Moreover, for severity of AKI, patients with preoperative serum albumin < 37.5 g/L tended to have a higher but not significant ratio in AKI stage 2 (2.6% vs 1.1%, P = 0.144) and a much higher ratio in AKI stage 3 (4.8% vs 0.7%, P < 0.001) than those with preoperative serum albumin ≥37.5 g/L.

To determine the cause of hypoalbuminemia, we further analyzed preoperative nutritional status using criteria of nutritional risk screening 2002 (NRS 2002). The results showed that patients with preoperative serum albumin < 37.5 g/L had significantly increased NRS score [4 (2, 4) vs 1 (1, 2), P < 0.001] and had a much higher ratio of NRS score ≥ 3 (77.5% vs 15.5%, P < 0.001).

Of all included patients, 14 patients (1.9%) died during hospital stay. Patients with preoperative serum albumin < 37.5 g/L had a mortality rate of 4.4%, which was much higher than 0.4% in patients with preoperative serum albumin ≥37.5 g/L (P < 0.001). The cumulative survival rate was also lower in patients with hypoalbuminemia (P = 0.003) (Fig. 2). Compared with that in non-AKI patients, the mortality rate was significantly higher in AKI patients (6.9% [13/188] vs. 0.2% [1/541]; P < 0.001). Kaplan-Meier analysis revealed that the cumulative survival rate decreased with increasing AKI severity (P < 0.001) (Fig. 3). In addition, postoperative AKI was associated with other worse outcomes, such as prolonged mechanical ventilation [53.4 (33.0, 73.8) vs 14.7 (11.1, 18.3) hours, P < 0.001], higher rate of other postoperative complications [0 (0, 2) vs 0 (0, 0), P < 0.001], ICU stay [4.0 (3.1, 4.9) vs 2.0 (1.8, 2.3) days, P < 0.001], postoperative hospital stay [17.8 (14.8, 20.9) vs 12.3 (11.3, 13.3) days, P < 0.001], and higher total cost [13,453 (8538, 20,228) vs 11,306 (6277, 16,400) dollars, P < 0.001] (Table 4, Additional file 2: Table S2). We also further analyzed AKI patients, preoperative hypoalbuminemia (< 37.5 g/L) was associated with more use of MV (72.4% [71/98] vs. 56.7% [51/90]; P = 0.024), longer ICU stay [4.5 (3.3, 5.7) vs 3.4 (2.1, 4.8) days, P = 0.027], higher occurrence of postoperative complications [1 (0, 3) vs 0 (0, 1), P < 0.001], and higher mortality (11.2% [11/98] vs. 2.2% [2/90]; P = 0.020) and total cost [15,160 (10,345, 22,221) vs 12,111 (6262, 17,763) dollars, P = 0.011] (Table 4). All the dataset of our study are available [see Additional file 7: Dataset].

Fig. 2
figure 2

Cumulative survival rate of patients with preoperative serum albumin < 37.5 or ≥ 37.5 g/L.

Full size image
Fig. 3
figure 3

Cumulative survival rate of patients with stages 1–, 2, and 3 AKI during postoperative hospital stay

Full size image
Table 4 Postoperative outcomes
Full size table

Discussion

Results of this retrospective study showed that preoperative hypoalbuminemia was independently associated with AKI occurrence in high-risk patients following non-cardiac surgery. In addition, more severe AKI stage was found in hypoalbuminemic patients. In accordance with previous reports of outcomes after non-cardiac surgery [1, 2, 20], the in-hospital mortality rate in AKI patients (6.9%) was very much higher than that in patients without AKI (0.2%). Other outcomes, such as ICU, postoperative hospital stay, and total cost, were also much worse in AKI patients. Furthermore, AKI patients with hypoalbuminemia had even more detrimental outcomes.

Cumulative evidence have shown that hypoalbuminemia is an important risk factor for postoperative AKI in various clinical settings [8,9,10,11,12]. However, in surgical settings, studies were mainly focused on cardiac surgery and transplant surgery [11, 12, 21,22,23]. Few studies have examined the effect of preoperative hypoalbuminemia on postoperative AKI patients undergoing non-cardiac surgery. Kim et al. [15] conducted a retrospective study enrolling 4718 patients who underwent partial or total gastrectomy for gastric cancer, and they revealed that patients with preoperative hypoalbuminemia, defined as < 40 g/L, had a significantly increased risk for AKI (OR 1.4; 95% CI 1.11–1.77). In patients following hip fracture surgery or total knee arthroplasty, after adjustment for confounders, early postoperative hypoalbuminemia has been shown to be strongly associated with AKI with a cutoff value of < 29 g/L and < 30 g/L [16, 17]. Recently, Kim et al. found that a preoperative serum albumin level < 38 g/L was independently associated with AKI (OR 2.465; CI 1.310–4.640) and mortality (OR 3.223; CI 1.959–5.305) in patients undergoing brain tumor surgery [18]. The finding from our study that preoperative hypoalbuminemia had a significant relationship with postoperative AKI was consistent with the results above. The cutoff value for hypoalbuminemia in our patients was 37.5 g/L, well above the usually accepted definition for hypoalbuminemia. Thus, our results suggested that with even a little decrease in preoperative serum albumin concentration, a higher incidence of postoperative AKI would occur in high-risk patients undergoing non-cardiac surgery.

Several possible mechanisms underlie this association. As a scavenger of radical oxygen species, combined with its anti-inflammation effects, albumin limits tubular cell apoptosis [24, 25]. Recent data have suggested that the integrity of the glycocalyx might be compromised in patients with hypoalbuminemia leading to loss of oncotic pressure gradients and barrier function, fluid leakage into the interstitium, and microvascular flow alterations [26, 27]. Moreover, ligation of endogenous toxin, modulation of nitric oxide and pharmacokinetic and pharmacodynamic effects of albumin also play an important role in renal protection [28, 29].

Albumin cutoff values vary between studies, and we attributed this difference to various study populations and types of surgery. In our study, as mentioned above, the cutoff value of 37.5 g/L had a sensitivity of 0.54, specificity of 0.67, and positive predictive value of 0.36; in patients undergoing brain tumor surgery, the cutoff value of 38 g/L had a similar sensitivity of 0.54, but lower specificity of 0.27 and positive predictive value of 0.04, which might be partly explained by the low incidence of AKI (1.8%) [18]. However, the AUC appeared to be similar with 0.624 in our non-cardiac surgery patients, 0.653 in hip fracture patients [16], and 0.684 in brain tumor patients [18]. Considering the possible negative association between serum albumin and AKI occurrence as reflected by research in patients undergoing cardiac surgery [11], we assumed that patients with a higher risk of postoperative AKI, such as having several comorbidities or undergoing general surgery [4], might have a lower tolerance threshold of serum albumin for AKI occurrence, thus requiring higher levels of serum albumin to protect perioperative renal function.

Currently, increasing amount of data revealed that postoperative AKI occurrence is associated with short-term adverse outcomes such as higher mortality and longer ICU and hospital stay [1, 2, 4, 11], which was also confirmed by our study. Furthermore, in AKI patients, preoperative hypoalbuminemia was associated with more use of MV, longer ICU stay, higher occurrence of postoperative complications, and higher mortality and total cost.

Unfortunately, there is still no effective treatment for AKI at present. Therefore, early recognition of high-risk patients and prevention of postoperative AKI become the first priority in clinical practice. Basic and clinical studies mentioned above indicated a potential benefit of correcting hypoalbuminemia for renal protection. Excitingly, Lee et al. [30] had made a step further. They recently performed a randomized controlled trial evaluating the effects of exogenous 20% human albumin solution vs saline on the incidence of postoperative AKI in adult patients with hypoalbuminemia (< 40 g/L) undergoing off-pump coronary artery bypass surgery. Their results have demonstrated that the incidence of postoperative AKI was lower in the intervention group than in the control group (17.6% vs 31.7%; P = 0.031). Multivariate logistic regression analysis revealed a renal-protective effect of albumin infusion with nearly 60% risk of AKI decreased (OR = 0.42, 95% CI: 0.21–0.83; P = 0.012). However, further studies are needed to address the results in the future, especially in patients undergoing non-cardiac surgery. Another way to increase preoperative serum albumin level is optimization of nutritional status. As shown in our study, 77.5% of hypoalbuminemic patients had preoperative NRS score ≥ 3, which indicated that malnutrition might be an important contributor to the occurrence of hypoalbuminemia. Until now, several studies have demonstrated significantly better results in overall and infectious complications in patients undergoing preoperative nutritional therapy [31,32,33]. However, data on the association between nutritional support and postoperative AKI were limited. Therefore, more work is needed to verify the effects of optimizing nutritional status on AKI, especially for patients undergoing non-cardiac surgery.

This study has major limitations. First, although we considered many perioperative AKI-related variables in our analysis, the effects of non-investigated factors could not be totally excluded. Second, given the lack of statistical power, subgroup analyses for the association of different preoperative albumin levels with AKI were not performed. Finally, in view of the retrospective and observational nature of this study, a causal relationship between preoperative hypoalbuminemia and risk of postoperative AKI could not be determined.

Conclusions

Our results showed that preoperative hypoalbuminemia was independently associated with AKI in high-risk patients following non-cardiac surgery, and postoperative AKI was associated with adverse prognosis. Prospective trials are needed to further identify the association between hypoalbuminemia and AKI and explore the potential beneficial effects of albumin infusion or specific nutritional therapy on postoperative AKI prevention.