Incidence and Associations of Acute Kidney Injury after General Thoracic Surgery: A System Review and Meta-Analysis

(1) Background: Acute kidney injury (AKI) is related to adverse outcomes in critical illness and cardiovascular surgery. In this study, a systematic literature review and meta-analysis was carried out to evaluate the incidence and associations of AKI as a postoperative complication of thoracic (including lung resection and esophageal) surgical procedures. (2) Methods: Adopting a systematic strategy, the electronic reference databases (PubMed, EMBASE, and Cochrane Library) were searched for articles researching postoperative renal outcomes that were diagnosed using RIFLE, AKIN or KDIGO consensus criteria in the context of a thoracic operation. A random-effects model was applied to estimate the incidence of AKI and, where reported, the pooled relative risk of mortality and non-renal complications after AKI. The meta-analysis is registered in PROSPERO under the number CRD42021274166. (3) Results: In total, 20 studies with information gathered from 34,826 patients after thoracic surgery were covered. Comprehensively, the incidence of AKI was estimated to be 8.8% (95% CI: 6.7–10.8%). A significant difference was found in the mortality of patients with and without AKI (RR = 2.93, 95% CI: 1.79–4.79, p < 0.001). Additionally, in patients experiencing AKI, cardiovascular and respiratory complications were more common (p = 0.01 and p < 0.001, respectively). (4) Conclusions: AKI is a common complication associated with adverse outcomes following general thoracic surgery. An important issue in perioperative care, AKI should be considered as a highly significant prognostic indicator and an attractive target for potential therapeutic interventions, especially in high-risk populations.


Introduction
With the advancement of medical science, the number of people diagnosed with pulmonary nodules or esophageal disease is steadily increasing [1,2]. Surgical tumor resection remains a mainstay of treatment and appears to be curative in most cases of primary cancer [3,4]. The practice of perioperative fluid restriction for the purpose of decreasing perioperative lung injury may increase the risk of renal insufficiency in patients undergoing thoracic surgery because of relative hypovolemia [3,5].
Acute kidney injury (AKI) is increasingly recognized as a common and serious complication after thoracic surgery, with incidence levels ranging from 2.2% to 35.3% [6][7][8], and it is related to long-term adverse consequences, including the progression to chronic kidney disease (CKD), as well as higher rates of cardiovascular disease and mortality [9][10][11][12]-even if renal function is clearly recovered at the time of discharge [5,8,12]. Hence, this might be significative of how, as a measurable signal of a prognostic indicator, postoperative AKI can be targeted for treatment.
In recent years, several meta-analyses associated with AKI have been performed in the context of cardiovascular and abdominal surgery [13][14][15][16][17][18]. The incidence of AKI after tho-racic surgery is second only to that for cardiovascular surgery and general surgery [17,19], but no systematic review has been conducted on this subject.
For this reason, a systematic review and meta-analysis was conducted in order to summarize the incidence of AKI in the context of general thoracic surgery and explore the relationship between AKI and other postoperative complications.

Materials and Methods
The meta-analysis was conducted in accordance with PRISMA and MOOSE guidelines [20,21], and the study followed the guidance of the STROBE evaluation method [22]. The protocol for this study is registered in PROSPERO under the number CRD42021274166.

Study Selection
Studies were included if they met the following eligibility criteria: (1) they were retrospective or prospective observational studies that provided data on the incidence of AKI after general thoracic surgery; (2) AKI was defined by any of the following consensus definitions: RIFLE (risk, injury failure, loss, end stage), AKIN (acute kidney injury network), or KDIGO (kidney disease improving global outcomes) [23][24][25]. Exclusion criteria included studies in which the diagnosis of AKI was based on diagnostic codes or the need for renal replacement therapy and studies in which cardiovascular surgery was reported.

Search Strategy
We electronically searched PubMed, EMBASE, and the Cochrane Library for original articles published from database inceptions in peer-reviewed journals excluding case studies, letters, reviews, minutes, and summary publications. The most recent search was performed on 11 June 2022. We used the following terms mapped to standard medical subject headings and free-text words for literature searches: AKI, acute kidney injury, renal failure, kidney failure, lung surgery, thoracic surgery, and esophageal surgery. Moreover, we manually reviewed the reference lists of all included studies and those from relevant reviews and meta-analyses in order to identify additional studies. The search was not limited by any restrictions on language. Supplementary Data S1 presents the detailed search strategy adapted for each database.
Two investigators (Y.Y. and S.X.) independently and sequentially reviewed the retrieved articles to determine their eligibility. Any disagreement between the review authors was resolved through consensus, or where necessary, by a third party (B.Y.) until a consensus was reached. EndNote 20.0 literature management software was used for the screening process.

Data Abstraction
The following information was obtained from the study using a structured data collection form: the title of the article, date of publication, country where the research was conducted, study design, diagnostic criteria for AKI, risk factors for AKI, incidence of AKI, incidence of severe AKI requiring renal replacement therapy (RRT), frequency of other postoperative complications (pulmonary complications including atelectasias, pneumonia, and respiratory failure, as well as cardiac complications including myocardial infarct, unstable arrhythmias, and congestive heart failure), length of hospital stay, and mortality (in-hospital mortality and 30-day mortality).

Study Quality
The quality of each study was independently assessed by two authors (B.Y. and H.Z.) using the Newcastle-Ottawa scale, which is one of the methods for measuring the quality of observational studies [26]. Any disagreement between the review authors was resolved through consensus, or where necessary, by a third party (X.T.). Studies were allocated a maximum of 9 points in value and defined as good (7)(8)(9), fair (4-6), or poor (0-3).

Statistical Analysis
Analyses were performed utilizing STATA version 16.0 (STATA Corp., College Station, TX, USA). Summary effect sizes were calculated as relative risks (RR) with 95% confidence intervals (95% CI) for dichotomous outcomes. To explain the presence of the possibility of between-study heterogeneity, data synthesis was conducted using a random-effects model, including the pooled effects of AKI in all studies as well as the relative risks of nonrenal complications in patients with AKI. The I 2 index (>50% indicating medium-to-high heterogeneity) and Cochran's Q test (p < 0.05 for statistical significance) were utilized to determine the between-study heterogeneity. We conducted sensitivity analyses as well as a funnel plot analysis and the Egger test for publication bias.

Results
In total, our search identified 1260 records of which 113 records were removed as duplicates and a further 1091 records were excluded for not meeting our eligibility criteria upon the screening of the title and abstract. After a full-text examination, 18 studies met our selection criteria [5][6][7][8][9][10][11][12][27][28][29][30][31][32][33][34][35][36]. The reference lists for the bibliographies of the text articles were screened, which revealed two reports that met our inclusion criteria [19,37]. Finally, 20 studies including 34,826 patients in all reporting AKI outcomes in the setting of thoracic surgery were included in our analysis. The study identification and selection procedures are presented in Figure 1. of observational studies [26]. Any disagreement between the review authors was resolved through consensus, or where necessary, by a third party (X.T.). Studies were allocated a maximum of 9 points in value and defined as good (7)(8)(9), fair (4-6), or poor (0-3).

Statistical Analysis
Analyses were performed utilizing STATA version 16.0 (STATA Corp., College Station, TX, USA). Summary effect sizes were calculated as relative risks (RR) with 95% confidence intervals (95% CI) for dichotomous outcomes. To explain the presence of the possibility of between-study heterogeneity, data synthesis was conducted using a randomeffects model, including the pooled effects of AKI in all studies as well as the relative risks of non-renal complications in patients with AKI. The I 2 index (>50% indicating mediumto-high heterogeneity) and Cochran's Q test (p < 0.05 for statistical significance) were utilized to determine the between-study heterogeneity. We conducted sensitivity analyses as well as a funnel plot analysis and the Egger test for publication bias.
A subgroup analysis was carried out to explore the effects of different diagnostic criteria and surgical sub-categories on the incidence of AKI. There were no significantly different results found between the first two subgroups listed (p = 0.136 and p = 0.203, respectively, Table 2), while huge heterogeneity existed between the studies overall. Risk factors potentially related to the development of AKI were examined in some studies, and patients with pre-existing renal disease were significantly more likely to develop AKI. Twelve studies excluded patients with hemodialysis preoperatively, among which the preoperative eGFR < 15 mL/min/1.73 m 2 [33,35,36]; only one study emphasized that the study population was derived from a cohort consisting of only patients with an eGFR ≥ 60 mL/min/1.73 m 2 [19]. Three studies enrolled patients with normal renal function [7] (normal preoperative SCr and BUN [28], or excluding SCr > 2 mg/dL [12]). However, we did not find a difference in the pooled incidence rate at 9.3% (4.8-13.8%) versus at 9.4% (6.8-12.0%) versus at 4.6% (0.9-8.2%) (p = 0.092, Table 2) among the five studies examining unselected patients [8,30,32,34,37] or among the studies excluding hemodialysis preoperatively [5,6,[9][10][11]19,27,29,31,33,35,36] and normal patients [7,12,28].
A meta-regression analysis of all the included studies demonstrated that the year of the study had no effect on the incidence of AKI (p = 0.162, I 2 = 98.15%, R2 = 3.06%) among patients after thoracic surgery.
There were five studies [8,11,27,29,32] that reported the rates of postoperative pulmonary complications and cardiac complications, among which we counted the one with the most cases. The pooled effect demonstrated the incidence of cardiovascular and respiratory complications was significantly higher in patients with AKI than in those without AKI (RR = 2.26, 95% CI: 1.22-4.16, p = 0.01, Figure S1; RR = 3.29, 95% CI: 2.48-4.35, p < 0.001, Figure S2, respectively). Similarly, nine studies [6,8,9,11,27,29,31,32,36] reported the length of postoperative stay, which demonstrated a significant increase in the amounts of patients with postoperative AKI. We did not pool the effect size due to statistical heterogeneity.

Evaluation for Publication Bias and Sensitivity Analyses
The validity of the use of funnel plots as a means for detecting publication bias is affected by small studies of low quality, which can cause funnel-plot asymmetry [38]. Small-study effects can be observed due to real differences and the publication bias is only one of the potential reasons for this [39]. There are other possible biases, such as the intensity of intervention and differences in the underlying risk that may also lead to funnel-

Evaluation for Publication Bias and Sensitivity Analyses
The validity of the use of funnel plots as a means for detecting publication bias is affected by small studies of low quality, which can cause funnel-plot asymmetry [38]. Smallstudy effects can be observed due to real differences and the publication bias is only one of the potential reasons for this [39]. There are other possible biases, such as the intensity of intervention and differences in the underlying risk that may also lead to funnel-plot asymmetry. Therefore, in addition to a funnel plot ( Figure S3), we conducted Egger's regression test for quantitative data and small-study effects. No significant publication bias was found in our meta-analysis, p = 0.209. We also performed a sensitivity analysis by sequentially removing each study, which showed a stable effect for each individual study ( Figure S4).

Discussion
In this meta-analysis, AKI was a relatively common complication in patients undergoing thoracic surgery with a pooled incidence rate of 8.8%, which is comparatively lower than for other types of surgery such as cardiac surgery (approximately 20-30%) [18], thoracic and abdominal aortic surgery (approximately 10-30%) [40], and major abdominal surgery (approximately 15%) [17]. In addition, the majority of patients with postoperative AKI had a case which was mild in severity, and less than 0.5% of patients in most studies experienced severe AKI requiring RRT [9,11,12,19,[29][30][31]35,36]. Where renal recovery after AKI was reported, the AKI had resolved at discharge in more than 70% of patients [5,8,12,30], and in the majority of patients, long-term renal function returned to normal [8]. However, we were unable to obtain a pooled analysis of the rates of renal recovery due to differing definitions and methods of assessment. Nevertheless, we did observe from the studies that quite a few patients who experienced AKI had ongoing renal dysfunction or even required renal replacement therapy [6,8,28,32].
Despite the low incidence, patients who experience AKI after thoracic surgery still experience prolonged hospital stays and increased mortality [6,8,9,11,27,29,31,32,36]. Even though the RIFLE classification has several important limitations, it may affect the early diagnosis and treatment of AKI and increase mortality [41,42]; we did find a significant difference in the mortality of patients with or without AKI when excluding the study adapting RIFLE criteria [5,6,8,9,11,28,29,31,32]. Regretfully, none of the studies reported a correlation between the duration of renal impairment and the risk of mortality post thoracic surgery. Only one showed that the mortality risk tended to become progressively greater with the advancing stages of AKI [27], and another one found a significant difference in the long-term mortality of patients [32]. Interestingly, Samuel et al. found that surgical procedures were related to a lower risk of death; we suspect that the results are attributable to the selection of a healthy patient with the highest benefit for selective surgery and the resolution of the self-limited conditions fixed to selective surgery [43]. Moreover, even slight AKI is still related to adverse events, such as the development of cardiovascular and pulmonary complications. Thus, perioperative prophylaxis and treatment remain critically important to the prevention of postoperative renal injury.
On the other hand, substantial heterogeneity was found in the rate of AKI, even though our analysis focused on studies using consensus AKI definitions. Contrary to our expectation, the incidence of AKI did not significantly differ in our subgroup analysis, and the heterogeneity amongst all the studies could not be partly explained by these factors. Moreover, we found no significant correlation between the year of the study and the incidence of AKI post thoracic surgery in meta-regression analysis. The incidence of perioperative AKI was basically unchanged despite significant advances in diagnosis, surgical techniques, and perioperative management. [44]. Finding promising therapeutic advances to diagnose and treat perioperative AKI is still a challenge.
The perioperative period is when special pathophysiological conditions challenge the diagnosis of AKI, such as muscle injury, volume overload or hypovolemia, and the release of aldosterone and vasopressin from stress [45,46], which may well have an impact on the measurements of creatinine levels and oliguria [35,36,47,48]. Doctors and researchers have striven to find a "troponin-equivalent" marker for the precise identification of patients with AKI at an early stage, such as cystatin C, kidney injury molecule-1, or neutrophil gelatinaseassociated lipocalin [49][50][51]. What is more, the urinary biomarkers tissue inhibitor of metalloproteinases-2 (TIMP-2) and insulin growth factor-binding protein 7 (IGFBP7) have been approved to help in identifying patients at a high risk for AKI in some interventional studies in cardiac and visceral surgery settings [52][53][54], as well as for improving the prediction levels of RRT and 30-day mortality after cardiac surgery [55]. The advantage is that they are able to detect kidney stress prior to injury or loss of function, allowing for much earlier therapy as compared to management guided by serum creatinine and urine output [54].
The mechanism of connection between surgery and AKI is still complex. The neuroendocrine response to hypotension, inflammation, and surgical trauma could probably damage kidney perfusion [29,34,56,57]. The reduction in renal blood flow and the reduction of renal oxygen supply lead to renal tissue hypoxia, which causes a cascade and further increases systemic inflammation [58,59]. AKI should be considered as a multi-organ system problem leading to dysfunction in the pulmonary, cardiac, neurologic, immunologic, and gastrointestinal systems, especially in high-risk groups [58].
There are conflicting data on the application of ACEIs/ARBs to AKI. ACEIs/ARBs have been commonly prohibited preoperatively to prevent intraoperative hypotension [49][50][51]60,61]. While some other researchers have argued that the beneficial pleiotropic effects of ACEIs/ARBs go far beyond blood pressure reduction, ACEIs/ARBs can improve renal recovery or reduce the fibrotic processes of renal function impairment after AKI, which is associated with a better prognosis for patients with AKI [62,63]. Quite a few studies have identified that preoperative hypoalbuminemia was independently associated with AKI [6,29,36]. Li et al. discussed how serum albumin may have some reno-protective effects in improving renal perfusion, binding endogenous toxins and nephrotoxic drugs, and scavenging reactive oxygen species [64]. Wiedermann et al. performed a meta-analysis that determined hypoalbuminemia was an independent predictor of AKI and AKI-related deaths [65]. As for the colloid infusion during surgery, there is no consensus [66], while hydroxyethyl starch solutions should be used with caution in high-risk patients undergoing thoracic surgery and have been discouraged in recent years [6,9,29].
Developing successful therapies to treat AKI has always been an elusive effort. Numerous agents (N-Acetylcysteine, the lipid-lowering 3-hydroxy-3-methylglutaryl coenzyme, dexmedetomidine, and so on) have shown promise [67][68][69], while clinical effectiveness varies between studies and the current evidence does not confirm the effectiveness of the agents in the treatment or prevention of AKI [30,70,71]. A single measure for prevention and therapy measures for AKI do not work well in clinical practice. A combination of treatments-including nutritional support [72] and glycemic control [25], minimizing nephrotoxic medication exposure [73], and hemodynamic optimization [29,34,74,75]-have largely been studied. Recently, new biomarkers of AKI have been discovered and validated [49][50][51]55]. In general, a combination of the biomarkers (urinary TIMP-2 and IGFBP7) for the early detection of perioperative kidney damage and accelerated intervention schemes seems to be the basis for AKI prophylaxis and treatment in surgical settings [52,55].
There are several limitations to our meta-analysis. Firstly, there are statistical heterogeneities in our meta-analysis. Subgroup analyses and meta-regression analysis cannot adequately explain the considerable heterogeneity. Secondly, AKI diagnosis in our analysis was mainly based on changes in serum creatinine and urine output, and there is limited data on novel biomarkers of AKI. Lastly, all the studies we included were observational studies. Thus, we merely demonstrated an association between AKI and increased complications after thoracic surgery, and the causality still needs to be confirmed by a large number of clinical trials or population-based studies of high quality.  [36] Lung Excluding hemodialysis elderly, hypertension, DM, use of ACEI/ARB, preoperative serum albumin and creatinine level, blood loss, intraoperative lowest MAP Abbreviations: ACEI/ARB, angiotensin-converting enzyme inhibitor/angiotensin II receptor blockers; AKI, acute kidney injury; ASA, American Society of Anesthesiologists; BMI, body mass index; CRP, C-reactive protein; DM, diabetes mellitus; FEV1, forced expiratory volume in 1 s; eGFR, estimated glomerular filtration rate; MAP, mean arterial pressure; NT-proBNP, N-terminal pro brain natriuretic peptide.
In summary, AKI is a common complication following general thoracic surgery and is associated with an increased risk of further non-renal postoperative complications and mortality. Despite the progress in perioperative management, the incidence of AKI in patients does not appear to have improved, suggesting the need for a greater attention to AKI following thoracic surgery, especially in particular populations.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jcm12010037/s1, Supplementary Data S1. Search terms for systematic review. Table S1. Newcastle-Ottawa Scale for assessing the quality of studies in metaanalysis. Table S2. Reported stage of postoperative AKI. Table S3. Rate of postoperative mortality in patients with/without postoperative AKI. Figure S1. Forest plot of cardiovascular complications in patients with/without postoperative AKI. Figure S2. Forest plot of respiratory complications in patients with/without postoperative AKI. Figure S3. Funnel plot from all trials. Figure S4. Sensitivity analysis by sequentially removing each study.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.