Abstract
ASSESSMENT OF KIDNEY FUNCTION IN PATIENTS WITH DECOMPENSATED CIRRHOSIS: DOES REMOVING RACE VARIABLE IMPROVE PERFORMANCE OF NEW GFR EQUATIONS?
Background:
Recently, new estimating GFR equations without race were proposed and rapidly implemented across the US (Inker NEJM 2021 CKD-EPI AS). Given the importance of kidney function assessment in cirrhosis/dual organ transplantation, we examined the performance of novel equations as compared to protocol measured GFR (reference standard) in patients listed for liver transplantation. In addition, we compared performance to equation developed in cirrhosis population (Asrani Hepatology 2020/2021 GRAIL).
Methods:
Measured GFR (mGFR) by iothalamate clearance at protocol time points before LT was used as reference. The Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI AS, Age, Sex, no race) equation was compared to CKD-EPI (2012), MDRD-4, MDRD-6 and Gfr Assessment In Liver disease (GRAIL) in the following subsets (1) cirrhosis (2) cirrhosis with ascites (3) cirrhosis with mGFR < 30 mL/min/1.73 m2 (4) black vs non-black race. We examined bias (difference between mGFR and eGFR, negative implying overestimation and positive implying underestimation), precision (width of confidence interval) and percent agreement within 30% of measured GFR (P30).
Results:
Between 1985 and 2015, there were 2,090 unique patients with measured GFR. As compared to the reference standard, GRAIL had lower bias (4.06 ml/min) followed by CKD-EPI AS 6.67ml/min (8.99 ml CKD EPI 2012; 11.75 ml/min MDRD 4 16.25 ml/min MDRD 6) across entire range of GFR as compared to the reference standard.(Figure)
Low GFR: For mGFR <30ml/min CKD-EPI AS overestimated GFR by 18 ml/min with low precision. This was higher than alternate equations. GRAIL had the smallest overestimation by 7ml/min (Table) For CKD-EPI AS, only 13% of measurements were within 30% of reference values (62% for GRAIL) and there was only 19% agreement between eGFR and mGFR CKD categories (49% for GRAIL). 64 percent of mGFR<30ml/min were incorrectly classified to a higher category by CKE-EPI AS.
Race: In blacks, CKD-EPI AS underestimated eGFR by 17.9 ml/min with a lower percent agreement between eGFR and mGFR categories as compared to other equations. This underestimation with the reference standard was almost 3 times as high as non-blacks (5.96ml/min) and twice as high as compared to CKD-EPI 2012 (7.6ml/min). Only 53% of blacks were categorized in appropriate GFR categories as compared to 66% with GRAIL. 12.6% percent mGFR>60 were classified as having worse kidney function (lower GFR stages) by CKD-EPI AS. In contrast, GRAIL underestimated by 3.8ml/min in blacks which was similar to non-blacks (4.75ml/min).
Conclusion:
Implementation of novel eGFR equations (CKD-EPI AS) without race may not accurately capture true kidney function in cirrhosis. In low GFR, it may not pick up kidney dysfunction. Among blacks, it may misclassify patients as having kidney dysfunction. Implementation needs to be tempered given relevant implications for management as well as decisions regarding dual organ transplantation.