Determinants of acute kidney injury after cardiac surgery: a systematic review

Acute kidney injury following cardiac surgery (CS-AKI) represents a severe postoperative complication, negatively impacting short-term and long-term mortality. Due to the lack of a specific treatment, effective prevention remains the most powerful tool to overcome the CS-AKI burden. Improving the preventive strategies is possible by establishing appropriate preoperative risk profiles. Various clinical models were proposed as a means to assist physicians in stratifying the risk of CS-AKI. However, these models are used for predicting severe forms of CS-AKI, while their predictive power for mild forms is insufficient. Our paper represents the first systematic approach to review all proposed preoperative risk factors and their predictive power. Our strategy is the starting point for selecting and comparing the predictive elements to be integrated into future risk models. Heart failure, chronic hyperglycemia, anemia, obesity, preoperative exposure to nephrotoxic drugs or contrast media, inflammation, proteinuria, and pre-existing kidney disease were systematically reviewed and were found to be associated with an increased risk of postoperative CS-AKI. As no externally validated and universally accepted risk models currently exist, the clinical judgment and a good knowledge of the preoperative risk factors in the light of new evidence may help personalize preoperative risk profiles as the cornerstone of prevention measures.


Introduction
Acute kidney injury (AKI) following open-heart surgery represents a severe postoperative complication, negatively impacting short-term and long-term mortality. Given the large number of patients undergoing cardiovascular surgery (CS), postoperative AKI incidence is high, ranging from 20% to 70% of cases (Vives et al., 2020). Severe forms of CS-AKI necessitating renal replacement therapy (RRT) have a lower prevalence (≤ 4%) but are subject to high mortality rates ranging from 40% to 70% (Nadim et al., 2018;Vives et al., 2020). However, increased mortality was reported in less severe CS-AKI forms, as an increase of 0.5 mg/dL in serum creatinine is linked to an almost 3-fold higher thirty-day mortality (Duchnowski et al., 2018;Lassnigg, 2004). Furthermore, a small postoperative increase in serum creatinine (0.3 mg/dL) is associated with a significantly higher long-term mortality and a 3-fold higher risk of developing end-stage renal disease (ESRD) (Rydén et al., 2014).
Various clinical models were proposed as a means to assist physicians in stratifying the risk of CS-AKI: Cleveland and modified Cleveland scores, Mehta score, and Simplified Renal Index (SRI) score (Huen and Parikh, 2012;O'Neal et al., 2016;Vives et al., 2020). However, these scores are used for predicting severe forms of CS-AKI, while their predictive power for mild forms is poor. There is an imperious need for improved strategies regarding the prevention, diagnosis, and treatment of AKI related to openheart surgery. Early recognition of AKI plays an important role, since the cornerstone of CS-AKI management is represented by preventive measures, as there is no specific treatment.
The objective of this updated systematic review is to weigh the impact of all reported risk factors in the development of CS-AKI and pave the way to more robust prevention strategies.

Study Selection
Studies involving ≥ ten patients over 18 years of age that reported original data regarding the preoperative risk factors for the development of AKI related to specific cardiac surgery types were eligible. Trials were included if adverse renal outcome was defined according to KDIGO, AKIN, or RIFLE criteria (Khwaja, 2012;Lopes and Jorge, 2013). Non-human studies, papers available only in abstract, letters, case reports, and meta-analysis were excluded.

Data Extraction
Data extracted from each included study were: authors' last name, year of publication, type of study, the preoperative risk factors being evaluated, number of patients, age median/mean, the proportion of patients exposed to a particular risk factor (when available), type of surgical intervention, criteria used to define CS-AKI, number of CS-AKI events. When available, data are presented as percentages, mean/median values, risk ratio (RR), odds ratio (OR), and range of variation. Also, when reported, the ability of a particular preoperative risk factor to improve the prediction power of a clinical model was documented.

Outcomes
We assessed the risk of developing CS-AKI in relation to the various preoperative risk factors reported in the included trials.

Quality Assessment
The quality of the non-randomized included trials was assessed using the Newcastle-Ottawa scale for cohort studies concerning three criteria (study group selection, the comparability of the study groups, and the evaluation of the outcome of interest) (Stang, 2010). The risk of bias for randomized trials was appraised using the revised Cochrane Risk of Bias tool (RoB2) (Sterne et al., 2019).

Results
Our database search retrieved 841 results. The study selection process is illustrated in Fig. 1. After removing titles and abstracts irrelevant to the topic (n = 756), duplicates (n = 5), meta-analysis (n = 4), and studies that did not meet the inclusion criteria (n = 53), 23 papers were included for the systematic review.
Regarding the metabolic risk factors, Hertzberg et al. (2015) found that both type 1 and type 2 diabetes mellitus were associated with an increased risk of postoperative CS-AKI independently of pre-existing kidney disease. CS-AKI risk was higher in type 1 diabetes mellitus (OR 4.89, 95% CI 3.82-6.25) than type 2 diabetes mellitus (OR 1.27, 95% CI 1.16-1.40). A high preoperative level of hemoglobin A1cis associated with a markedly increased risk of AKI after coronary artery bypass grafting (CABG) even in nondiabetic patients. Nonetheless, results have a limited impact due to the small number of patients and the low rate of CS-AKI events (n = 19, 9.4%). (Kocogulları et al., 2017). Similarly, Oezkur et al. (2015) reported that chronic hyperglycemia (hemoglobin A1c ≥ 6.0%) is associated with a higher risk of CS-AKI in patients undergoing CABG regardless of diabetic status. The study is limited by the small size of the patients cohort (n = 307). Obesity (body mass index ≥ 30 kg/m 2 ) was also linked to an increased risk of postoperative CS-AKI after multivariate adjustments (O'Sullivan et al., 2015).
Preoperative anemia is significantly associated with CS-AKI, according to De Santo et al. (2009) and Karkouti et al. (2011). Besides, Karkouti et al. (2011) showed that erythrocyte transfusion on the day of surgery also correlated with higher postoperative AKI rates. Moreover, the correlative effect had a greater magnitude in anemic patients than those without anemia (P-value for trend < 0.0001 vs. 0.1).
Data regarding preoperative treatment with ACEIs or ARBs is discordant. In a randomized trial, van Diepen et al. (2018) reported no difference in CS-AKI incidence between patients who continued the treatment with ACEI/ARB and those in whom treatment was discontinued two days before surgery (P = 0.991). Moreover, continued treatment with ACEI was not associated with higher intensive care unit length of stay (P = 0.420) or increased mechanical ventilation duration (P = 0.680). However, Coca et al. (2013) reported a higher risk of postoperative CS-AKI in patients who continued treatment with ACEI/ARB than those without a history of ACEI/ARB administration. Nonetheless, only patients exhibiting a high CS-AKI risk were enrolled, limiting the study's relevance regarding lower-risk patients. Unexpectedly, Barodka et al. (2011) reported a lower risk of CS-AKI in patients treated with ACEI/ARB. However, the small number of CS-AKI events (n = 19, 5.5%) is an essential limitation of the study.
Two studies (Jiang et al., 2018;Ranucci et al., 2013) assessing the impact of cardiac catheterization within seven days before surgery and coronary angiography on the same day as surgery reported an increased risk of postoperative CS-AKI. Moreover, Jiang et al. (2018) described an increased CS-AKI risk for a dose of contrast media (CM) > 240 mg/kg (OR 2.490, 95% CI 1.392-4.457). Additionally, the incidence of CS-AKI was higher when cardiac surgery was performed within seven days after contrast exposure (with a dose of CM > 240 mg/kg) compared to longer in-  2014) did not report a higher incidence of CS-AKI in patients who underwent coronary angiography within seven days before surgery (Borde et al., 2019) or in patients with prior cardiac catheterization (Özkaynak et al., 2014), respectively. However, the former study included only patients who underwent off-pump CABG, making it difficult to extrapolate the data to on-pump surgical techniques.
High-sensitivity C-reactive protein (hs-CRP), a marker of inflammation, was associated with CS-AKI. Han et al. (2017) reported that patients with higher hs-CRP levels exhibited a greater risk of CS-AKI after multivariate adjustments.
Pre-existing kidney disease represents an inherent risk factor for postoperative CS-AKI. Shlipak et al. (2011) reported that cystatin C had a stronger correlation with the risk of CS-AKI (OR 4.8, 95% CI 2.9-7.7) compared to creatinine or estimated glomeru-lar filtration rate (eGFR). Besides, only the worst kidney function quintile was independently linked to a greater risk of severe CS-AKI (OR 3.6, 95% CI 1.3-10.1 for cystatin C and OR 2.0, 95% CI 0.9-4.5 for creatinine, respectively). However, no significant correlation between eGFR quintile categories and severe CS-AKI was identified.
Galectin-3 is used as a cardiac biomarker in chronic kidney disease patients. Wyler von Ballmoos et al. (2018) reported that patients with higher levels of Galectin-3 had an approximately threefold higher risk of stage 2 or 3 CS-AKI and a two-fold more significant risk of stage 1 CS-AKI. Preoperative urine albumin to creatinine ratio (UACR) and dipstick proteinuria was associated with an increased risk of postoperative CS-AKI, as reported by  and Huang et al. (2011). However, most of the patients included in the former study experienced only mild CS-AKI, limiting the ability to assess a relationship between proteinuria and more severe CS-AKI. Notably, the latter study reported that more severe proteinuria had a higher risk of CS-AKI necessitating RRT (OR 7.29, 95% CI 3.00-17.73, P = 0.004). The quality of non-randomized studies was assessed using the Newcastle-Ottawa scale for cohort studies, and the results are summarized in Supplemental (Table 1). RoB2 assessed the risk of bias of the randomized trial (van Diepen et al., 2018), and the overall risk of bias was estimated as low.

Discussions
Our paper represents the first systematic approach to review all proposed preoperative risk factors and their predictive power.
Our strategy is the starting point for selecting and comparing the predictive elements to be integrated into future risk models. Systematically reviewing all risk factors is useful for both physicians and patients as the timely identification and correction (when possible) of preoperative risk profiles may result in safer surgical strategies and better postoperative outcomes.
CS-AKI represents a frequent complication without a specific treatment. Effective prevention, thus, remains the most powerful tool to overcome this problem. Improving the preventive strategies is possible by establishing appropriate preoperative risk profiles. Current clinical models display a low predictive value for milder forms of CS-AKI, although less severe forms are more common and have a negative impact on short and long-term mortality. Therefore, new and integrative models are needed to encompass risk predictions for all severity categories accurately. Before the construction of new models, the first essential step is to comprehensively review all updated original work studying preoperative risk factors for CS-AKI.
Management of pre-existing heart failure is vital as it is associated with a higher incidence of postoperative CS-AKI. Hertzberg et al. (2019) stratified the CS-AKI risk according to the left ventricular ejection fraction (EF). Patients with an EF lower than 30% had a greater risk of CS-AKI than those with a higher EF (OR 1.32, 95% CI 1.06-1.65). In addition, heart failure was associated with increased 30-day mortality. However, an important limitation of this study is that left ventricular filling pressures and relaxation were not considered. Also, patients with high BNP and NT-proBNP levels had a greater incidence of postoperative CS-AKI, as reported by Patel et al. (2012), andBelley-Côté et al. (2016). BNP levels enhanced the clinical model's predictive performance proposed by Patel et al. (2012). In addition, Belley-Côté et al.
A significant number of patients undergoing cardiac surgery have diabetes mellitus, another critical risk factor for CS-AKI. Hertzberg et al. (2015) showed that the therapeutic regimen influences CS-AKI risk in diabetic patients. A greater risk was reported in patients treated with insulin whether combined with oral antidiabetic drugs or not (OR 1.82, 95% CI 1.61-2.06) compared to those treated with oral antidiabetic medication only (OR 1.23, 95% CI 1.08-1.41) or diet only (OR 1.13, 95% CI 0.93-1.38). More-over, Kocogulları et al. (2017) and Oezkur et al. (2015) reported that chronic hyperglycemia (defined by a high preoperative level of hemoglobin A1c) was associated with postoperative CS-AKI, irrespective of a confirmed diagnosis of diabetes mellitus. This suggests that hemoglobin A1c levels may be useful in the preoperative risk stratification, even for non-diabetic patients. Nonetheless, higher hemoglobin A1c levels were not linked to an increased postoperative mortality or necessity of RRT. Furthermore, obesity represents a potentially modifiable risk factor independently associated with postoperative CS-AKI (O'Sullivan et al., 2015), pointing up the importance of body weight control.
De Santo et al. (2009) and Karkouti et al. (2011) reported that patients with anemia or erythrocyte transfusions had an increased incidence of postoperative CS-AKI. These results indicate that early correction of anemia may play an essential role in reducing the postoperative risk.
International consensus guidelines advocate for preoperative discontinuation of ACEIs and ARBs (Nadim et al., 2018), although data regarding their CS-AKI development risk is discordant. Coca et al. (2013) reported an increased risk of CS-AKI for patients treated with ACEIs or ARBs. Conversely, a randomized trial (van Diepen et al., 2018) revealed similar risks after medication discontinuation. Surprisingly, Barodka et al. (2011) revealed a protective effect of the preoperative treatment with ACEIs or ARBs against CS-AKI development. However, the latter study only included patients over 65 years, limiting the results' interpretation for the younger population.
The international consensus recommends no exposure to contrast agents 24 to 72 h before cardiac surgery (2C level of evidence) (Nadim et al., 2018), although studies are inconsistent. A singlecenter study from China (Jiang et al., 2018) reported that the dose of the contrast agent is also essential, suggesting that patients receiving a dose higher than 240 mg/kg should postpone the surgery (> 7 days). Ranucci et al. (2013) described an increased risk of postoperative CS-AKI in patients exposed to contrast agents on the surgery day. Conversely, two studies (Borde et al., 2019;Özkaynak et al., 2014) reported no influence of the contrast exposure on CS-AKI incidence. However, a subgroup analysis (Borde et al., 2019) showed that patients with pre-existing kidney disease (eGFR < 50 mL/min) had a greater risk of postoperative CS-AKI (OR 2.70, 95% CI 1.4-5.4, P = 0.005) when surgery was performed within seven days after coronary angiography. Thereby, a personalized approach that integrates the time difference between the two interventions, the contrast agents type, and dose, and the preoperative kidney function may prove to be more effective than the time interval strategy alone.
A South Korean observational study (Han et al., 2017) revealed that patients with high hs-CRP levels had an increased risk of postoperative CS-AKI after excluding septic patients (OR 1.88, 95% CI 1.392-2.525, P < 0.001). Moreover, a preoperative hs-CRP greater than 1 mg/dL was linked to a higher long-term mortality (HR 1.55, 95% CI 1.261-1.914, P < 0.001). High-sensitive CRP represents a cost-effective, routinely available biomarker for CS-AKI risk stratification in preoperative settings.
The preoperative kidney function is crucial as it indicates the baseline functional reserve and may identify the patients with a high risk of CS-AKI. Preoperative serum creatinine, cystatin C, and eGFR are used for evaluating kidney function. Mainly, cystatin C displays a stronger association with postoperative CS-AKI compared to the others (Shlipak et al., 2011). However, only serum creatinine and eGFR are routinely used due to lower costs.
Galectin-3 is a novel biomarker linked to chronic kidney disease, heart failure, and inflammation. Wyler von Ballmoos et al. (2018) reported that patients with a high concentration of Galectin-3 had a greater incidence of CS-AKI, and the addition of Galectin-3 improved the predictive power of the existing clinical models. Coca et al. (2012) and Huang et al. (2011) reported that UACR and dipstick proteinuria was associated with a higher risk of CS-AKI. However, there was no association between proteinuria and CS-AKI in patients with eGFR < 45 mL/min/1.73 m 2 . These results suggest that proteinuria should be incorporated in CS-AKI prediction clinical models. Dipstick proteinuria is an inexpensive alternative to UACR. Its cost-effectiveness may facilitate the integration in clinical risk models.
Regarding preoperative intra-aortic balloon pump (IABP) placement, the international consensus position (Nadim et al., 2018) is that IABP could have a beneficial effect on postoperative CS-AKI incidence. This is based on the results of a meta-analysis of 17 studies (Wang et al., 2016) reporting that the preoperative placement of IABP in patients at high risk was associated with a lower incidence of AKI after CABG (OR 0.54, 95% CI 0.36-0.79, P = 0.002).

Conclusions
Our study may be regarded as a source of analysis of recent studies on risk factors of AKI after cardiac surgery. AKI following cardiac surgery represents a frequent and severe complication. As there is no specific treatment, prevention measures that identify and minimize preoperative risk factors play a crucial role. Heart failure, chronic hyperglycemia, anemia, obesity, preoperative exposure to nephrotoxic drugs or contrast media, inflammation, proteinuria, and pre-existing kidney disease are associated with an increased risk of postoperative CS-AKI. To date, there are no externally validated or universally accepted clinical models to accurately predict postoperative CS-AKI occurrence, especially in non-severe clinical forms. For this reason, the clinical judgment and a good knowledge of the preoperative risk factors in the light of new evidence may help to personalize preoperative risk profiles as the cornerstone of prevention measures.