According to the results of multivariate logistic regression analysis, HT, DM, CRP, PCT, APTT, and PLT were identified as independent predictors of SA-AKI. Clinical predictive model based on these six factors can be used for predicting the occurrence of SA-AKI, and showed high predictive power.
There are several potential explanations for these results. The most important explanations for this is related to the pathogenesis of the disease condition. Endothelial cell (ECs) play a key role in the balance between pro-inflammatory and anti-inflammatory mediators, regulating vascular tone, barrier permeability and activation of coagulation. Increased importance of pathogenic mechanisms of SA-AKI is attributed to kidney damage, resulting in a complex interaction between inflammatory cascade activation, deranged coagulation pathways and immunologic mechanisms. This in turn led to microvascular dysfunction, endothelial damage, leukocyte/platelet activation with the formation of micro-thrombi, epithelial tubular cell injury and dysfunction. ECs lose functions and switch from a quiescent to an active state and then exacerbate inflammatory response [23–24, 27–28]. Abnormalities in coagulation with concomitant inflammatory processes were closely linked to the systemic generation of thrombin and consequent formation of microthrombi, contributing to organ dysfunction . Vascular injury and inflammation plays an important role in coagulation during sepsis. Simultaneously, a process called immunothrombosis, wherein platelet aggregation determines the release of cytokines and chemokines from platelet granules, leading to the recruitment of leukocytes and local inflammation . Inflammation and disturbances in coagulation are inseparably tied, each acting as a positive feedback for the activation of the other . Research on COVID-19 sepsis showed that during the occurrence of inflammatory reaction in the systemic organs, the microvascular system is shown to be damaged, resulting in abnormal activation of the coagulation system. This is pathologically manifested as systemic small vasculitis and extensive microthrombosis. Additionally, Professor Peng Zhiyong of the Central South Hospital and others conducted a retrospective analysis of 138 patients with COVID-19, and showed that the coagulation pathway was activated in severe COVID-19 patients due to cytokine storms, resulting in excessive consumption of coagulation factors and platelets . Excessive consumption of coagulation substrate caused coagulation dysfunction, and formed a vicious circle, which led to the development of severe COVID-19 to disseminated intravascular coagulation (DIC) or multiple organ failure (MOF). Secondly, Zhipeng Xu et al have conducted a retrospective analysis on the routine coagulation indexes of 138 patients with septic shock caused by intra-abdominal infection and the results revealed that the plasma APTT, PT and D-dimer levels on admission to the ICU were shown to be as significant risk factors of AKI . Chantalle E. Menard et al have shown that both prevalence and incidence of thrombocytopenia are high in septic shock patients. Thrombocytopenia is associated with poor prognosis. Our research showed that compared with sepsis patients without AKI, patients with AKI had more anomaly inspection indicators, and this is because they demonstrated greater disturbances in blood biochemistry (e.g. CRP, PCT, APTT, and PLT). Recent studies have focused on establishing clinical prediction models for SA-AKI based on coagulation markers, but these models did not combine important factors such as inflammation indicators, and did not develop a prediction nomogram in patients with SA-AKI [14, 33, 35–42], and so their prediction performance is considered insufficient. Many studies have focused on the mortality of sepsis instead of identifying high-risk patients with poor prognosis early and to intervene in a timely manner [43–45]. Further studies are warranted to evaluate the advancements in machine learning techniques. Based on the pathophysiology of SA-AKI, it is necessary to establish a clinical prediction model by combining inflammation coagulation indicators to predict SA-AKI.
However, there are several limitations that should be acknowledged in this study. Firstly, this was a single-center retrospective study, and validation of our model in other populations is required to more precisely evaluate the stability of the risk factors. Secondly, the present study was subjected to selection bias due to retrospective design. Due to limited sample size, the statistical power in comparative analyses was reduced, resulting in bias and coincidence. Some studies have shown that factors, such as IL-6, urine output and body mass index (BMI) [46–49], could influence the development and outcomes of SA-AKI might be ignored owing to the lack of complete medical records in some cases or the difficulty in data collection. Additionally, subgroup analysis of patients with mild AKI was not conducted owing to small sample size in this population. Furthermore, the time series analysis for the prediction risk of SA-AKI within the first 24 h of admission to the ICU was not conducted.
Despite these limitations, our study has several notable strengths. To the best of our knowledge, there were very few studies that has established the clinical prediction model of SA-AKI. Our study is the first to construct a nomogram for predicting SA-AKI based on clotting and inflammatory markers within 24 h of ICU admission. Although this was a single-center retrospective study, our research has important positive results and clinical significance. Traditional diagnostic indicators such as sensitivity, specificity and AUC can only measure the diagnostic accuracy of the predictive model, failing to consider the clinical utility of the specific model. Therefore, full advantage of DCA was taken and the clinical characteristics of patients were integrated into analysis. The calibration plot in our study showed excellent concordance between the nomogram prediction and the actual observations of SA-AKI, and the proposed nomogram model was shown to be clinically useful by DCA.
It is necessary to conduct further research on SA-AKI in order to identify high risk populations, employ preventive strategies, guarantee an early diagnosis and prompt and adequate treatment, which ultimately improved the patient outcomes. Our study could assist physicians in conducting early clinical interventions before the occurrence of AKI. Also sepsis stratification should be done in patients based on the severity of AKI. This method can form an intervention strategy based on the guidance of biomarkers to rationally use strained medical resources. Further external verification of our model should be conducted and the model should be displayed in the form of APP, web page or machine scoring to facilitate clinical practice.