Key findings
In this retrospective single-center study, we evaluated the preoperative, intraoperative, and ICU risk factors associated with CSA-AKI development. Overall, 20.8% of patients undergoing cardiac surgery were diagnosed with AKI. Of these, 15.5% were diagnosed upon hospital admission when the SC level at baseline was incorporated − all these patients were admitted for emergency or urgent cardiac surgery. Correction of the SC level for fluid balance increased the total number of AKI cases only marginally, and changed AKI staging in only six patients. In addition, we identified various factors associated with CSA-AKI development, whereas use of minimally invasive surgery, preoperative PASP, and volume of RBC transfusion were identified as potentially modifiable risk factors. Analyses of ROC curves had moderate-to-good discriminatory power for CSA-AKI, with a C-statistic ranging from 0.69 (preoperative variable) to 0.77 (preoperative and intraoperative variables). Finally, none of the collected variables were associated with renal recovery upon hospital discharge.
Relationship with previous studies
The prevalence of AKI in the total cohort and according to stage was similar to that in a meta-analysis involving > 320,000 patients undergoing cardiac surgery published recently [23]. Interestingly, correction of the SC level for fluid balance did not impact the diagnosis or staging of AKI significantly. This finding is in contrast with the work of Moore and colleagues, which showed an increase in AKI occurrence of ~ 12% when adjustment of fluid balance was made [24], and may reflect inter-hospital differences in intraoperative and postoperative standards of care of adult patients undergoing cardiac surgery. Although we detected significant differences in the total amount net fluid balance on postoperative day (POD)1 and POD2 between the AKI group and non-AKI group, these differences were markedly lower compared with those detailed by Moore and colleagues [24], and may explain the differences in AKI occurrence after adjustment of fluid balance.
As expected, we identified various preoperative and intraoperative variables associated with CSA-AKI development. Patients with increased susceptibility to kidney injury (e.g., those with long-term hypertension, DM, or higher age) have a substantially greater risk of developing AKI [25]. Similar to a recent study [26], we identified PSPAP as a risk factor for CSA-AKI. PSPAP has also been independently linked to increased mortality after valve surgery [27]. Increased PSPAP may be due to pulmonary hypertension secondary to left-heart disease but our ability to identify the prevalence and etiology of pulmonary hypertension in this study is limited. Also, the sensitivity and specificity of echocardiography to diagnose pulmonary hypertension are modest [28, 29]. Nevertheless, pulmonary hypertension and right-heart failure have emerged important risk factors for CSA-AKI, with venous congestion as the common cause [30, 31]. Hence, future mechanistic studies investigating therapies mitigating preoperative PSPAP may offer protection against CSA-AKI.
Also, we identified the use of minimally invasive surgery, which was the predominant surgical technique in our study population, to be associated with a low prevalence of AKI; this is likely because of the low risk of postoperative complications and blood transfusion requirement, lower risk of postoperative systemic immune inflammatory syndrome, accelerated recovery, and low use of potential nephrotoxic medications (e.g., non-steroidal anti-inflammatory drugs) [32]. Reports have indicated the noninferiority of this approach to conventional surgery [33], and recent guidelines recommend minimally invasive surgical procedures where expertise is available with an IIa class of recommendation (33). Studies in respect to renal outcome after minimally invasive surgery have focused on patients undergoing mitral valve or aortic valve surgery [32, 34, 35]. Murzi et al. [35] showed a 1% incidence of CSA-AKI after mitral valve surgery using video-assisted mini-thoracotomy, although in general CPB and cross-clamping times are longer in the port-access procedures [36]. The present study confirms the reported renoprotective benefits of minimally invasive surgery, and this approach may be considered as an alternative to standard median sternotomy valve surgery in patients at high risk for CSA-AKI.
In accordance with the literature [37, 38], we identified perioperative transfusion of packed RBCs to be associated with a high OR of developing CSA-AKI. Changes in intraoperative levels of hemoglobin can cause a proportional decrease in the oxygen-carrying capacity of the blood, which leads to impaired oxygen delivery and hypoxia in renal tissue, particularly if anemia is acute or severe [39]. RBC transfusion, however, has also been associated with increased morbidity and mortality, presumably due to infectious and non-infectious risks [40, 41]. Therefore, it is plausible that, by decreasing the incidence of blood transfusions and bleeding complications (e.g., with minimally invasive valve surgery), CSA-AKI and its associated complications may be reduced. Of note, both variables transfusion of packed RBCs and minimally invasive surgery were not collinear in our analysis emphasizing their independent role in the development of CSA-AKI.
The ability of the model to predict CSA-AKI was worse when compared with previously published and externally validated risk-prediction scores, with the C-statistic ranging from 0.81 to 0.85 [42, 43]. However, those studies also included postoperative variables, which may explain their better performance. Unexpectedly, we did not detect variables that influenced renal recovery.
Study implications
In line with other studies, our study implied that the pathophysiology of CSA-AKI is complex and current risk-prediction tools show only moderate-to-good calibration, suggesting significant heterogeneity in the underlying populations. However, among potentially modifiable risk factors, use of minimally invasive surgery and therapies mitigating preoperative pulmonary arterial pressure and intraoperative blood loss may offer protection against CSA-AKI.
Strengths and limitations of our study
No other research teams have utilized full KDIGO consensus criteria for AKI to incorporate the SC level at baseline and correction of the SC level after surgery for fluid balance for the diagnosis and staging of AKI. The other strength of our study was the assessment of multiple variables at baseline, intraoperatively, and in the ICU. In addition, only few studies evaluated the impact of minimally invasive surgery on CSA-AKI.
Study limitations were its retrospective design, relatively small number of patients, and its single-center location. Thus, our results cannot be extrapolated directly to other patient populations. Also, due to the retrospective nature of the study, complete elimination of selection bias is unlikely. Multivariate logistic regression analyses verified with respect to collinearity and the interactions between variables were performed to minimize the differences in measured confounders; however, unmeasured confounders cannot be accounted for.