In this study, we found that the incidence of CA-AKI was greater under the new 2020 diagnostic criteria. Our risk model, which used only three risk factors (emergency procedures, SCr, and proteinuria) showed a good discrimination and significant calibration in the validation dataset. An increasing risk score was strongly associated with the development of CA-AKI as well as in-hospital and post-discharge hemodialysis. Our risk model may serve as an easy to apply guide for evaluating the individual patient risk for the development of CA-AKI after the PCI.
CA-AKI accounts for roughly 10% of hospital-acquired renal failure but the incidence of CA-AKI depends on the definition used. Formerly, CA-AKI had been defined as an increase of > 0.5 mg/dL or > 25% in the baseline SCr within 48–72 h after exposure to contrast media4. The newer guideline issued in 2020 lowered the threshold for an absolute increase in the SCr, but increased the threshold for a relative increase, as well as limiting the time period to 48 hours, making it unclear whether the new criteria would lead to an increase or decrease in the incidence of CA-AKI by definition. In 2020, the American College of Radiology and the National Kidney Foundation issued a new definition of an absolute (0.3 mg/dL within 48 hours after contrast exposure) or relative (> 50%) increase in the baseline SCr, and this can be a standard reference for defining both the onset and severity of CA-AKI15. In our study, the incidence of CA-AKI increased significantly from that under the old definition, 1.9%, to that under the new definition, 5.0%. In a recent report, the optimal definition of CA-AKI in predicting adverse cardiovascular outcomes and post-PCI mortality was compared according to four different definitions: an absolute elevation in the SCr of ≥ 0.3 mg/dL or ≥ 0.5 mg/dL at 48 hours post PCI or a relative elevation of ≥ 25% or ≥ 50%19. They found that an absolute elevation of ≥ 0.3 mg/dL in the SCr 48 hours post PCI predicted the outcomes most accurately. The new definition may have increased the sensitivity of CA-AKI diagnosis and led to a rise in the cardiologists' awareness of CA-AKI.
A majority of the prior studies reported that mild-to-moderate renal dysfunction was a risk factor for CA-AKI8–10, 18. In agreement, the SCr was an independent predictor of CA-AKI in our study. Emergency procedure was also associated with CA-AKI, consistent with the previous reports, probably due to vasoconstriction or a reduced renal blood flow associated with a sudden hemodynamic deterioration caused by acute coronary syndrome. In our study, the contrast dose was not independently associated with an increased risk of CA-AKI in the logistic analysis. That was probably because the operators performed the PCI with a low amount of contrast given the severity of the renal function shown in Supplementary Table 2. Our observation that the contrast media was not an independent predictor of CA-AKI in the patients undergoing a PCI was in agreement with the recent literature9, 10, 16.
The main strength of our risk model was that it stratified the patients at risk of CA-AKI well using only 3 pre-procedural factors. We could explore the useful risk score using inexpensive and routinely collected measurements in the preprocedural setting. We have shown that our risk score achieved a similar predictive value as compared to Mehran’s score 8 and Inohara’s score9. Mehran’s risk score, which has been widely used from 2004, includes 8 factors: 6 pre-procedural factors (age > 75 years, congestive heart failure, anemia, diabetes, hypotension, and eGFR) and 2 procedural factors (intra-aortic balloon pump and the contrast media volume). Inohara’s 7 risk score factors are all pre-procedural: age, heart failure, diabetes, previous PCI, hypertension, SCr, and acute coronary syndrome. A simpler, intuitive, and easily obtainable risk score may be useful given the high morbidity of CA-AKI and the importance of early detection. Recently, Allen et al. examined the discrimination and calibration of the risk prediction models for CA-AKI accompanying cardiac catheterization in a meta-analysis5. They found that CA-AKI prediction models had a good discrimination (C-statistic 0.78), but had high heterogeneity (I2 statistic = 95.8%, Cochran Q-statistic P < 0.001), partly due to the differences in the CA-AKI definitions used in each study. They further reported that models that included postprocedure in addition to preprocedure variables did not show significantly higher c-statistics than the models that used only preprocedure variables. A preprocedural risk stratification provides cardiologists with an opportunity to better understand the risks of CA-AKI before embarking on a PCI, providing for kidney protective strategies such as preprocedural hydration and minimization of contrast volume.
In this study we showed that proteinuria was independently predictive of CA-AKI in patients undergoing a PCI. Our results were supported by a previous study, which reported an incremental value in the proteinuria for predicting CA-AKI after cardiac catheterization14. While proteinuria is thought to be a manifestation of glomerular damage, recent experimental data has shown that albumin, which accounts for most of the urinary protein content, upregulates the expression of pro-inflammatory and profibrotic mediators in cultured renal tubular cells20. These results suggest that ultrafiltered albumin secreted from renal tubular cells damages the renal function via complement activation or an inflammatory process leading to interstitial fibrosis and tubular damage. Patients with proteinuria may have an impaired physiological adaptability and be less tolerant of renal hemodynamic changes and nephrotoxic injuries such as from contrast media.
This was a retrospective and single-center study that excluded patients undergoing emergency cardiac surgery and patients who died within 2 days after the procedure. Such patients may have a high prevalence of renal dysfunction or various comorbidities. Thus, we likely underestimated the incidence of CA-AKI. However, the number of patients excluded for any reason was 27 out of 1,281. Secondly, we evaluated our model using an internal validation dataset. Thirdly, we determined the severity of the proteinuria using the dipstick method because it allows for an easy and rapid diagnosis. The dipstick method, however, evaluates gross albuminemia (urinary albumin excretion > 300 mg/day) rather than microalbuminuria (30–300 mg/day). In a future study, we need to test the effect of microalbuminemia in assessing the risk of CA-AKI. In addition, the dipstick test can give false negative results21.