Perils of Estimating GFR in Patients with Cirrhosis

In my attempt to highlight nephrology content in non-renal journals, I will this month focus on a series of articles in Hepatology concerning eGFR in cirrhotic patients. We are all aware of the concerns using creatinine-based tools for estimating GFR in this patient cohort given their malnourished, low muscle-mass state. The use of Cystatin-C based equations may theoretically be more informative as this protein is not influenced to the same degree by non-renal factors. However, Cystatin-C is far from perfect and is influenced by sepsis, inflammation and steroid use. Other concerns include the use of the MDRD-6 equation due to the inaccurate determination of albumin (may receive IV infusions) and urea (increased by GI bleed/steroids) in these equations.

As liver patients with renal dysfunction have such a poor prognosis, they are prioritized for liver transplant by way of inclusion in the MELD score, the prognostic tool used to allocate liver allografts. Since adopting MELD, the number of combined liver-kidney transplants (cLKT) has continued to grow, with cLKT considered when GFR<30mls/min and sometimes at higher eGFR. As the demand for kidneys continues to outstrip supply, nephrologists in particular have legitimate concerns regarding the potential for inappropriate use of precious renal allografts in cLKT. This may occur if eGFR underestimates true GFR as may occur with creatinine-based equations. Also concerning is overestimation of true GFR resulting in denial of cLKT where it may actually be indicated, leading to inferior patient outcomes. (See the post by Andrew regarding combined liver-kidney transplant allocation).

Francoz et al studied 300 patients with cirrhosis being evaluated for transplant that had iohexol clearance measured. This was compared to MDRD-4, MDRD-6 and CKD-EPI equations (all using creatinine only) and found that MDRD-6 was the most accurate although it did underestimate true GFR and all 3 equations had poor correlation (R2 0.37-0.4). MDRD-4 and CKD-EPI overestimated GFR especially in those with GFR<60mls/m.

DeSouza et al also looked at patients being evaluated for transplant (n=202) and measured GFR using inulin clearance. Throughout all severities of cirrhosis, Cystatin-C equations were superior with CKD-EPI (Cys-C) the best, compared to creatinine-based MDRD & CKD-EPI methods (which significantly over-estimated true GFR). Of note it outperformed CKD-EPI (creatinine-cystatin-C combined).

Mindikoglu et al examined 72 outpatients with stable cirrhosis comparing CKD-EPI (creatinine-cystatin-C combined) to 24-hour creatinine clearance, Cockcroft-Gault equation and multiple other creatinine-based methods including MDRD & CKD-EPI. Their gold standard was iothalamate clearance. CKD-EPI (creatinine-cystatin-C) performed better than all others including CKD-EPI (Cys-C), unlike DeSouza et al.

Confusing right? What we can take away from these useful studies is that Cystatin-C based equations may be better than creatinine alone equations (remember Francoz et al did NOT use Cystatin-C). However, it should be noted that the diagnostic performance of the best equations in the studies was still markedly lower than reported in validated normal populations. 
My feeling is that in borderline cases when a cLKT is being considered, we need additional data. I would consider a borderline case stable renal dysfunction in the eGFR 20-50mls/min range (arbitrary I know!), not including co-existent ESLD/ESRD or obvious acute hepatorenal cases which will recover with a functioning liver allograft. As renal biopsy is usually not desirable in patient with chronic liver disease, it seems sensible to actually measure GFR in these cases. This appears to be the prudent approach to take to strike a balance between providing a kidney to those who need it and not inadvertently denying an organ to a wait-listed ESRD patient.