Contrast-associated acute kidney injury (CA-AKI) is a common complication after coronary angiography that is closely related to adverse clinical prognoses such as cardiovascular disease, renal events, and long-term mortality [1‐3]. CA-AKI is usually defined as an increase in serum creatinine (SCr) ≥ 50% or ≥ 0.3 mg/dl within 72 h after exposure to a contrast agent, but the definition of CA-AKI is not limited to this. It is defined as an increase in SCr by 0.5 mg/dl or 25% within 72 h after using a contrast agent, which has also been confirmed in many clinical studies. In fact, the incidence and prognostic value of CA-AKI are different with different definitions [4‐8]. It has been found that different definitions have different efficacies in predicting long-term mortality in STEMI patients, and the criteria of the Acute Kidney Injury Network (AKIN) has better efficacy [4]. Different definitions also have different performances in terms of population-attributable risk (PAR) in AMI patients; the standard definition, which is an increase in SCr ≥ 50% or ≥ 0.3 mg/dl within 72, has the highest PAR [6]. In previous studies, diabetes mellitus was considered a risk factor for CA-AKI, and it has been applied to several high-efficacy predictive models for CA-AKI [9‐11]. However, to date, there are few articles comparing the association between different definitions and long-term prognosis in diabetic patients.

Therefore, we evaluated the association between different definitions of CA-AKI and long-term mortality in diabetic patients after coronary angiography in this article.

We analyzed the relationship between different definitions of CA-AKI and long-term prognosis in diabetic patients and compared the PARs of long-term mortality. According to our findings, the incidence of CA-AKI_A was slightly lower than that of CA-AKI_B, while the mortality rate was the opposite. In addition, we also found that the mortality rate of patients with CA-AKI was higher than that of patients without CA-AKI. After adjusting for demography, drugs and cardiorenal risk factors, an increased risk of long-term mortality was associated with CA-AKI_A and CA-AKI_B. However, CA-AKI_A was associated with higher risk of long-term mortality and PAR than CA-AKI_B.

Without a uniform definition, the incidence of CA-AKI was between 8.49 and 12.91% in patients with diabetes mellitus undergoing coronary angiography. The incidence of CA-AKI in patients with diabetes indicated in Chalikias's review was similar [20]. Our results suggest that regardless of how CA-AKI is defined, it leads to a poor long-term prognosis. In a study discussing the clinical significance of CA-AKI, CA-AKI was associated with death compared with non-CA-AKI, even when different definitions of CA-AKI were applied [5]. This further confirmed those results. On the other hand, in a study with a median follow-up period of 365 ± 130 days, Centola et al. compared the effects of different definitions of CA-AKI on long-term death in 402 patients with STEMI and found that each different definition increased the risk of death [4]. CA-AKI may be an important risk factor for long-term mortality, rather than an intermediate, which is similar to Mehran's recently described viewpoint [21]. The research of Lassnigg et al. also supports this view, and they found that a small increase or decrease in serum creatinine was significantly related to death [22]. Fluctuation in serum creatinine may be due to unstable hemodynamics, and the downstream adverse events have little relevance.

As far as we know, few studies have compared the relationship between different definitions of CA-AKI and long-term mortality in diabetic patients. However, we evaluated the correlation between the two definitions of CA-AKI and long-term mortality in diabetic patients by cox regression analysis and PAR calculation. The significance of population attributable risk (PAR) is the proportion of cases that would not happen in the population without risk factors. Chen et al. also found that the 1^st most common definition of CA-AKI (≥ 0.3 mg/dl or ≥ 50%) and the 2^nd most common definition of CA-AKI (≥ 0.5 mg/dl or ≥ 25%) have different predictive powers for long-term mortality risk, and the 1st most common definition of CA-AKI is more closely related to all-cause death [5]. Recently, Lei et al. found that three different definitions of CA-AKI were also inconsistent in predicting death in AMI patients. Their conclusion was that the highest PAR was found in CA-AKI_A (a SCr elevation ≥ 50% or ≥ 0.3 mg/dl within 72 h), which is consistent with our results [6]. Benoit Guillon and his colleagues showed that the CA-AKI defined by SCr absolute increase ≥ 0.3 mg/dl or ≥ 50% over baseline was strongly correlated with 6 months all-cause death in patients with CAG [7]. And Gabriele Pesarini et al. found that an absolute increase of SCr > 0.3 mg/dl seems to be the most clinically meaningful cut-off value for CA-AKI and continuous renal damage monitoring [23]. These studies have shown that different definitions of CA-AKI have different strengths of association with death, and CA-AKI defined as SCr elevation ≥ 50% or ≥ 0.3 mg/dl within 72 h has a stronger association with long-term mortality. This further confirmed our results. Our results show that the HR and PAR of CA-AKI_A were higher than those of CA-AKI_B, which indicates that the CA-AKI_A definition is more closely related to long-term mortality. On the one hand, this may be related to the patient we included, and diabetic patients belong to high-risk groups. In a long-term hyperglycemic environment, HbAlc increases, oxygenated hemoglobin decreases, and microvascular perfusion is insufficient, which leads to kidney ischemia and hypoxia and aggravated kidney injury. On the other hand, when the "restrictive" criteria are adopted (an increase ≥ 0.5 mg/dl in SCr above baseline), patients with slight increase in serum creatinine may be ignored, which will have an impact on long-term death. When the "relativity" criteria (an increase in SCr > 25% above baseline) are adopted, the recognition ability will be reduced [24]. For example, in Gajewska's research, it was found that the renal damage caused by SCr increased from 1.2 to 1.6 mg/dl was much higher than that caused by SCr increased from 3.0 to 3.75 mg/dl [25]. In addition, in previous studies, mild increases in serum creatinine in patients with lower baseline serum creatinine are more likely to lead to significant deterioration of renal function. The majority of diabetic patients often have renal insufficiency, and a slight increase in "restrictive" criteria (≥ 0.3 mg/dl or ≥ 0.5 mg/dl) often cannot truly reflect the occurrence of kidney injury in diabetic patients. The "relativity" criteria are often more able to reflect the exact kidney damage based on the poor baseline renal function.

After adjusting for AMI, hypotension and HS-CRP, we found that CA-AKI was significantly associated with long-term mortality. AMI, hypotension and HS-CRP are the research hotspots of CA-AKI and prognosis. The association between AMI and CA-AKI has been confirmed in previous studies [26, 27], and Zafrir et al. also found that patients with AMI can have significantly increased long-term mortality [28]. Hypotension, as a strong predictor of CA-AKI, is also a good predictor of death. Compared with the risk found in our long-term follow-up, Chong et al. found that patients with hypotension had a high risk of death at 6 months [29]. As for HS-CRP, it is an important indicator of inflammation. In Xiao-Sheng Guo's et al. study, the inflammatory response was found to be an important mechanism for the occurrence of CA-AKI, and the risk of long-term mortality was also increased with an HS-CRP level > 7.3 mg/l [30]. The reliability of our analysis can be further enhanced by adjusting for variables of clinical significance.

At present, there were various definitions of CA-AKI [20], which makes clinicians confused about the diagnosis of CA-AKI. We confirmed that CA-AKI is an independent risk factor for long-term mortality in patients with diabetes, and the definition of CA-AKI_A is more suitable for diabetic patients than CA-AKI_B for assessing the relationship between CA-AKI and prognosis. At the same time, the strict definition of CA-AKI can prompt a poor prognosis. This is helpful for clinicians to identify and manage high-risk CA-AKI patients that are highly related to long-term death, which improving the prognosis. We found that among patients with diabetes, the definition of CA-AKI_A is more suitable for assessing the relationship between disease and prognosis, and more attention and care should be given to CA-AKI_A patients in clinical practice. The comparison of the definition is helpful for medical staff to identify the patients with CA-AKI who are most related to long-term adverse prognosis in high-risk groups (such as heart failure, chronic kidney disease, anemia, or diabetes), so as to carry out accurate management. This method of using cox regression analysis and PAR for comparison of definitions in high-risk patients to guide clinical management is worthy of promotion. Our study did not evaluate the impact of the progression of renal function on long-term mortality due to the lack of SCr during follow-up. In the study of Nemoto et al., patients with CA-AKI were more likely to suffer persistent impairment of renal function, which strongly affects long-term prognosis [31]. This may affect our results and has become a major limitation of our research.

Our research has several limitations. First, the patients we included in the study were diabetic patients, which will reduce the generalizability of the findings. Second, our study was only a sub-study of a single-center prospective observational study in South China, and the prevalence of CA-AKI was not significant. However, we are the first to explore the relationship between the definition of CA-AKI and long-term prognosis in diabetic patients, and we also found that our results were similar to those reported in previous studies involving high-risk groups. Third, we conclude that preventing the occurrence of CA-AKI can reduce the risk of long-term mortality but cannot directly confirm this conclusion. However, our study evaluated the association between CA-AKI and long-term death by calculating PAR and further verified the results of the multivariate analysis. Fourth, due to the limitation of the sample size of our study, the variables for multivariable analysis adjustment were limited. However, long-term mortality (all-cause mortality) was multifactorial and complicated. In addition, most of the patients included in our study had coronary artery disease. During follow-up, they may be more prone to cardiovascular mortality. This long-term mortality (all-cause mortality) is more close to cardiovascular mortality. Fifth, we lacked the SCr data of patients during the follow-up and were unable to assess the progression of renal insufficiency, which may have an impact on our results. However, we also adjusted for clinically significant variables related to long-term mortality in the multivariate analysis, which makes our results more credible. Future studies still need to supplement the assessment of patients' renal function and the endpoint of cardiovascular mortality to improve the credibility of the results.

Our conclusion proves that CA-AKI is an independent risk factor for the long-term death of diabetic patients, and the definition of CA-AKI_A is more suitable for diabetic patients. This is helpful for clinical patients to identify CA-AKI patients who are highly related to long-term death among diabetic patients, so as to give more kidney protection measures to improve their prognosis.