We read the report by Inohara et al that recently appeared in The American Journal of Cardiology . The investigators developed a preprocedural risk model for contrast-induced acute kidney injury (CI-AKI) in patients who underwent percutaneous coronary intervention (PCI) and demonstrated that adding procedural variables to the model does not significantly changes discrimination (C-statistics improving from 0.789 to 0.799) . With this study, the authors support our previous efforts aimed at identifying preprocedural variables associated with the risk of CI-AKI in patients who underwent primary PCI. It is crucial, indeed, to have a meaningful risk stratification at baseline that proves to be independent from subsequent procedural variables, which hardly can be anticipated at clinical presentation.
Two points deserve in our view a specific comment. First, in the study by Inohara et al, the use of radial access, compared with femoral, was significantly associated with a reduced incidence of CI-AKI (radial approach was used in 24% of patients experiencing CI-AKI, compared with 38% in patients not experiencing, p <0.001) . Although these results were not confirmed in the multivariable analysis, one may argue that the lower prevalence of radial access (37% vs 63%) in the derivation cohort might justify the lack of statistical significance to support inclusion of radial access in the model. Indeed, these data are consistent with a recent meta-analysis of observational studies in which radial approach was associated with a 48% relative risk reduction in the risk of CI-AKI (p <0.001). We expect to have more definitive conclusion about this issue from the Acute Kidney Injury-Minimizing Adverse hemorrhagic events by TRansradial access site and systemic Implementation of angioX study , in which the relation between vascular access and CI-AKI will be explored in a large population of patients with acute coronary syndromes randomized to radial or femoral access.
Second, although the contrast dose was slightly higher in the group experiencing CI-AKI, Inohara et al have not found contrast to be included in their models . As expected, patients experiencing CI-AKI had an almost twofold higher prevalence of overt renal dysfunction (indicated by creatinine >1.0 mg/dl in 44% vs 22% of patients). We have recently demonstrated that contrast volume remains a significant predictor of CI-AKI in addition to preprocedural variables only whenever normalized for renal function. To this purpose, the use of Net Reclassification Improvement analysis might be the only statistical technique to demonstrate the incremental discriminative power obtained with the addition of procedural variables to the baseline risk model. Further studies are needed to demonstrate whether a contrast volume threshold, adjusted for baseline renal function, might be ever considered a variable to be neglected in the prediction of CI-AKI. In other words and agreeing with the investigator’s considerations , it is likely that any operator-dependent action aimed at maintaining contrast volume as low as reasonably possible still has a positive effect in reducing the occurrence of CI-AKI even on top of baseline risk assessment. Quoted data from the large National Cardiovascular Data Registry, including about a million PCI procedures, indirectly confirm this issue.
In conclusion, Inohara et al should be congratulated for having added a significant piece of information to the body of knowledge about baseline indicators of the occurrence of CI-AKI after PCI. Regardless whether the preprocedural model is simple or complex, it should be prospectively tested in combination with variables under the control of the physician as an utmost opportunity to improve patient’s care.