The Aftermath of AKI: Recurrent AKI, Acute Kidney Disease, and CKD Progression

• diabetes mellitus
• acute kidney injury
• acute kidney disease
• chronic kidney disease

AKI and CKD are interconnected syndromes with significant long-term sequelae such as progressive CKD, cardiovascular events, and death.^1,2 Diabetes is not only a major cause of CKD, but is also an independent risk factor for AKI. AKI in diabetic patients is associated with medications such as renin-angiotensin system (RAS) blockade agents, sodium-glucose cotransporter-2 inhibitors, and shared risk factors such as hypertension and cardiovascular diseases.³ However, the actual estimates of AKI incidence in diabetic patients remains imprecise because most epidemiologic studies are on the basis of administrative codes rather than laboratory measurements of change in serum creatinine.⁴

In this issue, Hapca et al.⁵ investigated risk of AKI and the relationship between AKI occurrence and subsequent CKD progression in patients with diabetes. They utilized an observational cohort study, enrolling >18,000 Scottish participants with longitudinal data linked to local health care records and registry, and compared the incidence of AKI, subsequent decline in eGFR, and CKD progression in diabetic patients who were adequately matched with nondiabetic controls. They developed a novel algorithm to identify AKI episodes with the addition of serum creatinine up to 1 year after discharge. This approach facilitated the determination of the duration of the AKI episode and whether AKI progressed to acute kidney disease (AKD). However, a major limitation in such kidney registry studies is using creatinine both in the definition of exposure (AKI) and outcome (eGFR for development of CKD). In addition, the rate at which renal function worsens may not necessarily be quick enough to meet the definition of AKI but may rather manifest as AKD followed by continuous worsening of renal function to meet the definition of CKD. Also, AKI episodes in diabetic patients may require a prolonged recovery time with persistently elevated creatinine level before progression to CKD. With availability of longitudinal creatinine values, the authors were careful about misclassifying recurrent AKI and AKD states, after initial AKI episodes, as occurrence of CKD. The inclusion of post-AKI creatinine increased the detection of AKI from 28,306 to 40,567 cases. During the 8-year follow-up, patients with diabetes were almost five times more likely to experience AKI and more recurrent AKI than nondiabetic controls (121 versus 25 per 1000 person-years, adjusted rate ratio 4.7, 95% confidence interval 4.4 to 5.5), and those who had diabetes and CKD at baseline were two times more likely to experience AKI and more recurrent AKI compared with nondiabetic CKD controls (adjusted rate ratio 2.1, 95% confidence interval 2.1 to 2.5). However, there was no difference in the severity and duration of AKI between diabetic versus nondiabetic groups. Patients in both diabetic and nondiabetic groups who developed AKI had steeper eGFR declines before AKI episodes than those who did not develop AKI, although the decline was much more prominent in diabetic patients (slope pre-AKI versus no AKI −1.14 in diabetic group and −0.29 in nondiabetic group). Furthermore, AKI accentuated eGFR decline in both groups, and the effect was more prominent and statistically significant in nondiabetic patients (slope post-AKI versus pre-AKI −0.29 in diabetic group and −0.55 in nondiabetic group). The increased risk of kidney disease progression in diabetic patients after AKI episodes was similarly seen in the landmark Assessment, Serial Evaluation, and Subsequent Sequelae in AKI cohort in which patients had lower eGFR at baseline and were recruited in the acute hospital setting.⁶

As the authors discuss, the higher risks of AKI in diabetic patients may be associated with preexisting subclinical injury, such as glomerulosclerosis, interstitial inflammation, fibrosis, and compensatory glomerular hyperfiltration, either from diabetes or a higher burden of comorbidities.⁷ More importantly, the study highlights the effect of AKI on eGFR decline in both diabetic and nondiabetic patients. The casual relationship of AKI and the development of CKD is well established in epidemiologic studies.² In addition to eGFR, the presence of microalbuminuria after AKI is strongly predictive of transition to CKD and CKD progression.⁶ Multiple pathophysiologic pathways have been studied extensively to solve the conundrum of AKI to CKD transition. Inflammation, microvascular rarefaction, mitochondrial dysfunction, and cell senescence could eventually lead to maladaptive repair, tubular atrophy, and interstitial fibrosis.⁸ Therefore, it is not surprising that the urinary kidney injury molecule-1, TNF receptor-1, and basic fibroblast growth factor were shown to predict incident CKD or CKD progression in patients who developed AKI after cardiac surgery.⁹ In addition, persistent activation of the intrinsic RAS after prolonged ischemia was found to correlate with interstitial fibrosis in mice models and CKD progression in a small group of patients with acute tubular injury.¹⁰ Although RAS blockade or sodium-glucose cotransporter-2 inhibitor may decrease albuminuria, counteract overactive RAS, and provide long-term kidney benefit, the use of these therapies after AKI episodes may also increase the risk of recurrent AKI via hemodynamic changes. These episodes of recurrent AKI require careful scrutiny to minimize discontinuation of these renoprotective medications. However, these two important medications are not evaluated in this study and deserve further investigation.

This study demonstrated a robust approach to improve the detection of AKI in future large-scale epidemiologic research involving registries with serial measurements of kidney function. Although this algorithm does not allow early detection and prevention of AKI, this approach would assist future research in large cohorts or electronic health records to understand different AKI and AKD phenotypes, and thus fill the clinical gap of events leading to CKD. However, we do need to interpret this study with caution as comorbidities, the use of RAS blockade, and microalbuminuria are not adjusted and may attenuate the high risk ratio for AKI seen in diabetic patients. Additional analysis on how different stages or durations of AKI may affect long-term eGFR decline would also enrich existing knowledge. It is unclear whether these AKI episodes truly reflect intrinsic kidney injury or simply reflect fluctuations of serum creatinine during initiation or titration of certain medications, such as RAS blockade and diuretics, which are both commonly used in diabetic patients. In addition, diabetic patients underwent more frequent creatinine measurement, making it more likely for AKI to be captured, compared with nondiabetic patients, although sensitivity analysis comparing stage 2 and 3 AKI in diabetic and nondiabetic patients revealed similar results.

Prospective studies and clinical trials are needed to address these limitations, and to improve our knowledge of AKI events in diabetes and our approaches to reduce AKI to CKD transition. The National Institute of Diabetes and Digestive and Kidney Diseases has prioritized research in this area and will support multicenter consortium to conduct interventional trials to improve post-AKI care.¹¹ Prospective clinical trials could follow patients after hospitalization for AKI, use biomarkers to identify high-risk patients for CKD progression, and eventually explore the benefits of several interventions in these high-risk patients.

• Copyright © 2021 by the American Society of Nephrology