Assessing the risk of CKD progression using Kidney Failure Risk Equations

 “The best way to predict your future is to create it". Abraham Lincoln

Predicting the future is a tenuous concept and at face value is perhaps better suited to popular media and science fiction than medicine. We have an innate preoccupation with the future engendered since childhood and further developed throughout our early adult years. This is not an idle curiosity; at the crux of it is our desire to predict and plan for future challenges. It is this that is at the core of attempts to predict the future in healthcare; identifying risk factors for disease, identifying individuals at risk of disease progression and attempting to enable decision making for treatment in healthcare systems already under considerable service and financial demands.

The pioneering Framingham study was the benchmark study for risk prediction models in healthcare. Chronic kidney disease represents an equally significant health and economic burden, however it has not previously proved amenable to risk prediction models. Current risk stratification in CKD relies on eGFR. Unfortunately, use of eGFR may overestimate the severity of disease in CKD. In addition, eGFR decline may be non-linear and episodes of acute kidney injury may unpredictably shorten the time to ESKD.

CKD has represented a difficult challenge in creation of adequate risk prediction models. CKD is not a uniform entity—for many patients the presence of CKD may have no appreciable impact on their life whereas others will ultimately progress to ESKD. Timely and appropriate specialist referral is imperative to allow interventions and enable planning for renal replacement therapy. Scores that underestimate disease severity may result in inadequate preparation time resulting in patients “crash landing”, missed opportunities regarding slowing progression and multi-disciplinary team input. Overestimation of severity results in increased referrals to Nephrologists with a knock-on effect of increased waiting times, increased demands on the system, unnecessary patient anxiety and a potential delay in identification and treatment of those who will ultimately progress to kidney failure.

Prediction scores to determine which patients with CKD are likely to progress to ESKD would allow us to have patient-specific discussions regarding prognosis and would allow us to collaborate with primary care to select patients who need expedient specialist review. Given the importance of better ESKD risk progression, KDIGO Guidelines on CKD management recommend the use of risk models. The Kidney Failure Risk Equation (KFRE) has been shown to accurately predict the risk of progression in 2 large Canadian and 1 European cohort. 

• Tangri et al developed multiple models of kidney failure risk in a cohort from Toronto and validated the equations a cohort in Vancouver incorporating clinical and laboratory data of patients with CKD 3-5 referred to nephrologists over a 7 year period. The abbreviated KFRE consists of four variables (age, sex, eGFR and albumin-to-creatinine ratio), and the full eight variable KFRE includes calcium, phosphate, bicarbonate and albumin. The eight variable KFRE showed modest gains compared with the 4-variable model. Unsurprisingly, the ESRD risk was much lower in those who had CKD stage 3 at baseline (3%) compared to those who were already at CKD stage 5 at baseline (60%). About 40% had diabetes and 90% had hypertension. Early criticisms of the KFRE involved missing data in both the derivation and validation cohorts, and the cohort selected were already under the care of the nephrologist.
• The KFRE was then assessed in a multi-national cohort encompassing 31 cohorts of 721,357 patients with CKD 3-5 in more than 30 countries spanning 4 continents. This meta-analysis showed that original risk equations achieved excellent discrimination, however a calibration factor was required in the non-American cohort.  
• Peeters et al  applied the KFRE to a cohort in the Netherlands and externally validated the full and abbreviated forms KFRE in an independent CKD population.
• A further study by Tangri et al. published in AJKD involved use of a version of the 8-variable equation encompassing dynamic values for prediction; this involved longitudinal follow-up of a cohort from a Nephrology clinic over several years. 11% of the study population developed ESKD; unsurprisingly this was predominately those who had CKD 5 at baseline (60%). Change in GFR emerged as a strong factor in predicting those progressing to ESKD emphasizing the importance of monitoring change over time. This study will require further validation in an external cohort as in the previous study, however the dynamic approach is promising as it is able to incorporate information from future events with time dependent covariates throughout the follow-up period.

Obtaining the expected risk progression scores is now as simple as entering data into the QXMD online calculator or application. The possible implications for our future practice is promising, however there are further considerations. The equation could be used to guide referrals to Nephrology services as suggested by Tangri at his recent talk at the KDIGO Controversies Conference on Dialysis Initiation, Modality Choice & Prescription. The equation is yet to be validated in a primary care setting where the vast majority of patients with CKD are not referred for nephrology care. The full KFRE also requires additional data that is unlikely to be available in the primary care setting. ESKD is not the only important clinical outcome in CKD to be considered as there is an additional need to incorporate cardiovascular outcomes and mortality attributed to CKD. An alternative potential application of the KFRE is applying it to new patient referrals to Nephrology clinics. Applying the KFRE at first visit could give Nephrologists added confidence in both discharging patients and identifying those at high risk of progression who need to begin education and planning. The equation is well on the road to being useful in clinical practice and represents a potential game-changer in CKD management. Incorporation of novel biomarkers of CKD into existing scores may represent a future direction for predicting risk of progression. We await further studies testing the KFRE in the broader CKD population with excitement and anticipation.

Post by Laura Slattery, NSMC Intern 2018