Chronic Kidney Disease in Liver Transplant Candidates: A Rising Burden Impacting Post–Liver Transplant Outcomes

Patients

All patients listed for LT in the United Network for Organ Sharing (UNOS)/Organ Procurement and Transplantation Network (OPTN) registry between January 1, 2002, through December 31, 2017, were included in this study. Patients who were <18 years old, listed as status 1, who received exception points, or who underwent a living donor liver transplantation (LDLT) were excluded. We excluded those listed with exceptions because we believe that they are intrinsically different than those without exceptions with a different natural history of disease and therefore risk for CKD.

Renal Function

Serum creatinine and estimated glomerular filtration rate (eGFR) were determined longitudinally from the time of listing for LT to removal from the LT waiting list. Renal function was assessed every 7 days or longer. When a patient had more than 1 serum creatinine value for the same 7-day period, the first test result was used. We calculated eGFR using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) creatinine-based equation.13 We chose this equation because, of the glomerular filtration rate calculators that can be used with the data available in the UNOS/OPTN registry, the CKD-EPI equation most closely estimates glomerular filtration rate relative to glomerular filtration rate as measured by iothalamate clearance in patients with cirrhosis.14-17 Those on hemodialysis (HD) were treated as having an eGFR <15 mL/minute.

Because of the nature of the UNOS database, we compared measurements for AKI within 7 days and not 48 hours as used in current definitions.20

Renal Function Pattern in Those Listed <90 Days

Because we could not know whether patients listed for <90 days had CKD, no patient listed for <90 days was considered to have CKD, unless they were on HD for ≥42 days. That being said, those patients listed for <90 days were eligible to meet the criteria for AKI.

Covariates

Data were obtained from the UNOS/OPTN registry as of April 6, 2018. Demographic data were collected at listing. The following data were collected at listing and at the end of follow-up: total bilirubin, international normalized ratio (INR), presence of hepatic encephalopathy (HE), and presence of ascites. Cutoffs deemed to be implausible were as follows: total bilirubin ≤0 mg/dL, INR ≤0, and creatinine ≤0 mg/dL.21 Observations with implausible values were set as missing for that specific value. The Model for End-Stage Liver Disease–sodium (MELD-Na) score22 was calculated and capped at 6 and 40, per current liver allocation policy. Because the MELD-Na score was not implemented until 2016, the MELD score was used in the descriptive statistics and analysis of the development of CKD.23, 24 Listing diagnoses were grouped into the following common diagnostic categories: hepatitis C virus (HCV), nonalcoholic fatty liver disease (NAFLD; including cryptogenic cirrhosis and nonalcoholic steatohepatitis [NASH]), alcohol-related liver disease (ALD), autoimmune etiologies (including primary biliary cirrhosis, primary sclerosing cholangitis, and autoimmune hepatitis), and other. As has been done in previous studies,21, 25 regions were grouped according to median MELD score at time of LT into low (regions 3, 6, 10, and 11), medium (regions 2, 4, and 8), and high (regions 1, 5, 7, and 9) MELD regions. Donor characteristics included those used to calculate the donor risk index (DRI).26 The data were categorized into 4-year increments (2002-2005, 2006-2009, 2010-2013, and 2014-2017) based on the year of last follow-up to allow for an adequate sample size by year and to address changes in transplant policy (eg, implementation of MELD-Na and Share 35). The year of last follow-up was chosen to best determine the incidence of CKD at the time point closest to transplant.

Outcomes

The primary independent variable was the presence of CKD at the time of LT. A primary outcome of post-LT mortality was used to determine the impact of pre-LT CKD on post-LT survival. For this analysis, follow-up began on the date of LT and ended at the time of death or last update to the UNOS/OPTN registry.

Statistical Analysis

Population Characteristics

A total of 78,640 patients listed for LT met study inclusion criteria (Fig. 1). Of the 78,640 patients listed for LT, 39,719 (51%) received a DDLT, 20,791 (26%) either died on or were removed from the waiting list for “sickness,” and 18,130 (23%) were still waiting on the waiting list. Overall, the cohort consisted of 29,846 women (38%); 57,609 (73%) Caucasians; 18,589 (24%) patients with ALD; 27,289 (35%) patients with HCV; and 16,974 (22%) patients with NASH. Median age of the cohort at listing was 55 years (interquartile range [IQR], 48-60 years); median listing eGFR was 72 mL/minute (IQR, 43-98 mL/minute); median final eGFR was 57 mL/minute (IQR, 23-91 mL/minute); median listing MELD was 18 (IQR, 13-25); and median final MELD was 23 (16-32).

Among 39,719 DDLT recipients, 6269 (16%) patients met the CKD criteria at the time of last transplant (Table 1). Patients who developed CKD prior to LT were significantly older (58 versus 54 years; P < 0.001), more likely to be female (43% versus 34%; P < 0.001), and more likely to have a listing diagnosis of NAFLD/NASH (31% versus 21%; P < 0.001), ascites (55% versus 53%; P < 0.001), and diabetes mellitus (33% versus 20%; P < 0.001). They were also more likely to be from a high-MELD region (37% versus 32%; P < 0.001). Those who developed CKD had a significantly lower eGFR at listing (52 versus 72 mL/minute; P < 0.001; Table 1). During the study period, there were significant changes in the characteristics of patients undergoing LT. The proportion of patients meeting CKD criteria at the time of last follow-up increased from 7.8% in 2002 to 14.6% in 2017 (test for trend, P < 0.001; Fig. 2).

Cox Regression Analysis for Post-LT Survival

Of the 39,719 patients who underwent LT during the study period, 10,830 (27%) died after LT at a median of 2.3 years (IQR, 0.4-6.0 years). There was a significant decrease in both 1-year (12.4% in 2002 to 6.9% in 2016) and 5-year (23.9% in 2002 to 18.3% in 2012) post-LT mortality by year of transplant during the study period (P < 0.001 for both by test for trend). As compared with those without CKD, those with CKD prior to LT were significantly more likely to die after LT: 1-year mortality, 11.9% versus 9.0% (P < 0.001); and 5-year mortality, 21.5% versus 17.5% (P < 0.001).

In the univariate analysis for post-LT mortality, CKD at transplant was significantly associated with post-LT mortality (as compared with no CKD): stage 3 CKD, hazard ratio (HR), 1.16 (95% CI, 1.08-1.25); stage 4 CKD, HR, 1.42 (95% CI, 1.28-1.58); and stage 5 CKD, HR, 1.42 (95% CI, 1.31-1.54; Table 2).

Using this model, we discovered a number of additional findings. First, we found that any stage of CKD was associated with higher post-LT mortality, with the HR increasing at higher stages (as compared with no CKD): stage 3 CKD, HR, 1.11 (95% CI, 1.03-1.20); stage 4 CKD, HR, 1.19 (95% CI, 1.06-1.33); and stage 5 CKD, HR, 1.21 (95% CI, 1.11-1.32). Second, if instead of the diagnosis of CKD, the total days with an eGFR <60 mL/minute was incorporated into the final adjusted model, we found that for every 90 days with CKD, the risk of post-LT mortality increased by 2.7% (adjusted hazard ratio [aHR], 1.03; 95% CI, 1.02-1.04). Third, despite the significant improvement in LT outcomes by era, there was no significant interaction between the effect of pre-LT CKD on post-LT survival by era (P > 0.05 for all eras), meaning that we have not decreased the impact of pre-LT CKD on post-LT outcomes.

Mediating the Influence of Receiving an SLKT on post-LT Mortality in Those With CKD

Next, we wanted to better understand the impact of receiving an SLKT on post-LT outcomes, particularly in the 3736 (9%) patients with CKD who met the SLKT criteria. Of those with pre-LT CKD, 48.6% received an SLKT versus 6.4% of those without (P < 0.001). There were significant differences between those who did and did not receive an SLKT (Supporting Table 1). The Kaplan-Meier failure plot demonstrates that those with any chronic renal dysfunction (eg, met SLKT without SLKT, SLKT without meeting SLKT, or met SLKT and underwent SLKT) had significantly higher rates of post-LT mortality (P < 0.001 by log-rank test; Fig. 3). Additional multivariate models then demonstrated that there was no significant mediating influence of SLKT on the effect pre-LT CKD had on post-LT survival: 1-year proportion of the attributable effect of SLKT (4.0%; bias corrected 95% CI, −41.1% to 49.1%); 5-year proportion of the attributable effect of SLKT (6.0%, bias corrected −9.7% to 23.1%).

Impact of the Renal Function Pattern on Post-LT Survival

Of the 39,719 DDLT recipients, 5739 (14%) patients had AKI, 5188 (13%) patients had CKD, and 1019 (3%) patients had AKI on CKD at the time of their transplant. There was a significant trend for all renal dysfunction patterns (ie, AKI, CKD, and AKI on CKD) to increase during the study period (test for trend, P < 0.001). As compared with those patients with normal renal function, the rates of post-LT mortality were significantly higher in those with CKD (31% versus 26%; P < 0.001), AKI (29% versus 26%; P < 0.001), and AKI on CKD (31% versus 26%; P < 0.001).

In the univariate analysis, patients with CKD (HR, 1.33; 95% CI, 1.26-1.41), AKI (HR, 1.35; 95% CI, 1.27-1.42), and AKI on CKD (HR, 1.54; 95% CI, 1.37-1.73) had significantly higher risk of post-LT mortality as compared with patients with normal renal function at the time of LT. Similarly, in the final multivariate model after adjustment for age, MELD at transplant, etiology of cirrhosis, encephalopathy, diabetes at transplant, race, and era of transplant, each of the renal dysfunction patterns had greater post-LT mortality as compared with normal: AKI (aHR, 1.21; 95% CI, 1.13-1.27); CKD (aHR, 1.19; 95% CI, 1.12-1.27); and AKI on CKD (aHR, 1.28; 95% CI, 1.13-1.45).

That being said, there was no significant difference between the AKI or AKI on CKD groups and the CKD group (P > 0.05 for both comparisons).