The estimation of glomerular filtration rate (GFR) has relied upon the use of mathematical equations primarily based upon the measurement of serum creatinine and other parameters such as age, body weight, gender, and ethnicity. In 2009, the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) published a new equation that uses similar variables as the MDRD Study equation but, as shown in recent studies, estimates GFR more precisely. In a recent article in the American Journal of Kidney Diseases, Dr. Matsushita and colleagues from Johns Hopkins University, University of Alberta, Tufts Medical Center, and University of Calgary compare the CKD-EPI equation to the MDRD Study equation in a large cohort of patients in Canada. Dr. Kunihiro Matsushita (KM) from Johns Hopkins Bloomberg School of Public Health discusses the study with Dr. Matthew Sparks (eAJKD), eAJKD advisory board member.
eAJKD: Why are you interested in this topic and why is this important to study?
KM: Two years ago, we investigated this issue in the Atherosclerosis Risk in Communities (ARIC) Study. The CKD-EPI equation more appropriately categorized individuals with respect to long-term clinical risk compared with the MDRD equation, suggesting improved clinical usefulness in this middle-aged bi-ethnic population in the United States. However, there are some areas that the ARIC study could not explore. For example, does the CKD-EPI equation better classify individuals to help us predict adverse outcomes in the clinical setting? We were fortunate to collaborate with investigators from the Alberta Kidney Disease Network to address this topic.
eAJKD: Can you give us a brief description of the Alberta study patient cohort that was used in your study?
KM: The Alberta Kidney Disease Network (AKDN) is a collaborative group of nephrology researchers in the province of Alberta, Canada, with a mission to undertake clinical research and offer research training in kidney disease. AKDN identifies patients from laboratory data and then collates other available data sources. Data available in the AKDN include socio-demographic variables as well as clinical information such as comorbidities, health care costs, and several outcomes including mortality, myocardial infarction, and end-stage renal disease.
eAJKD: Can you tell us about the original cohort of patients that provided the CKD-EPI equation for eGFR?
KM: The original paper of the CKD-EPI equation was published in 2009 in the Annals of Internal Medicine. The equation was developed and validated in approximately 12,000 individuals with both measured and estimated GFR from 26 studies. These 26 studies consist of patients with chronic kidney disease (CKD) or living donors of kidney transplantation. Compared to the database that was used to develop the MDRD Study equation, a greater number of people were included in the CKD-EPI development and the range of measured GFR was broader. Given these facts, the CKD-EPI equation could estimate GFR more precisely in the higher ranges of kidney function.
eAJKD: An interesting part of this paper is the reclassification of patients to higher and lower eGFRs. Can you tell me who was more likely to have a higher eGFR using the CKD-EPI equation than with the MDRD equation? Was this appropriate?
KM: We observed in the AKDN as well as the ARIC study that younger individuals, women, and whites are more likely to be reclassified to a higher eGFR category by the CKD-EPI equation. There are more people who are reclassified upward compared to those who are reclassified downward. This is appropriate in the sense that the MDRD Study is known to underestimate GFR in some populations. We observed in both AKDN and the ARIC study that those who are reclassified to a higher eGFR category actually had lower risk of clinical outcomes. A strength of the AKDN data set is that given its large sample size, we could observe significantly better reclassification even after adjusting for potential confounders.
eAJKD: What about those who were reclassified to a lower eGFR?
KM: We observed only 1.2% of individuals who were reclassified to a lower eGFR category. These individuals have unique characteristics as compared to those who are not reclassified. They tend to be older with a mean age of 81 years, compared to a mean age of 50 years who stayed in the same eGFR category. The reclassification was consistent with clinical outcomes. Those who were reclassified downward actually had higher risk of clinical outcomes, even after accounting for potential confounders including age or traditional risk factors. We think these findings suggest that the reclassification by the CKD-EPI equation is clinically appropriate in both directions.
eAJKD: Can you discuss some of the limitations of the study?
KM: One of the main limitations is that the cohort mainly consists of whites. This means that our findings may not be necessarily generalizable to other ethnicities. Also, this data set does not have information on some important potential confounders, such as smoking, body mass index, or measured blood pressure. In addition, the follow up time was relatively short with a median of 2.8 years, limiting any long term conclusions from this study. Nevertheless, in ARIC we found that reclassification was associated with these clinically appropriate outcomes even in the long term.
eAJKD: Should the CKD-EPI equation be used routinely to estimate eGFR?
KM: Our results support the use of the CKD-EPI equation for eGFR reporting, particularly since the CKD-EPI equation uses identical variables with the MDRD Study equation and does not require additional laboratory measurements. From another perspective, if clinical laboratories do not change from the MDRD Study to the CKD-EPI equation, our results suggest that a substantial number of individuals will be incorrectly identified for being at higher risk, potentially leading to excessive tests and care. Given these facts, our findings suggest switching from the MDRD Study equation to the CKD-EPI equation in clinical practice.
(Note: there is a recent article on a related topic that appeared in JAMA.)