Comanagement With Nephrologist Care Is Associated With Fewer Cardiovascular Events Among Liver Transplant Recipients With Chronic Kidney Disease

Supplemental Digital Content is available in the text.


INTRODUCTION
Chronic kidney disease (CKD) is a common complication after liver transplant (LT), with an incidence of stage 4 or 5 CKD (defined as estimated glomerular filtration rate [eGFR], <30 mL/min per 1.73 m 2 ) up to 18% at 5 y after LT. 1 Many factors have been associated with increased risk of CKD post-LT, including pretransplant renal dysfunction, diabetes, hypertension, and immunosuppression with calcineurin inhibitors. 1,2 CKD is a well-known risk factor for cardiovascular (CV) disease in the general population. 3 Among LT recipients (LTRs), CV disease is a leading cause of mortality, and renal dysfunction at the time of and immediately posttransplant is associated with higher risk of major CV events. 4,5 In addition, other factors that have been associated with increased mortality after LT include pre-LT CKD, post-LT acute renal failure, and post-LT CKD. [6][7][8] Given the significant CKD burden in this patient population and the known negative effects of this condition on clinical outcomes, improvement in the management of CKD could have significant potential to improve clinical outcomes for LTRs.
In the general CKD population, prior data have shown that early nephrology referral for management of CKD is associated with lower hospitalizations and mortality rates. 9 Consequently, current clinical practice guidelines for the general CKD population recommend nephrology referral for several clinical scenarios. 10 These guidelines also provide recommendations on management of both CKD and many other chronic conditions that frequently occur in patients with CKD including hypertension, hyperlipidemia, diabetes, anemia, etc. Although the aforementioned clinical practice guidelines help guide management of patients with CKD, published guidance on the management of CKD among LTRs has focused on adapting key aspects from the nontransplant CKD guidelines that are most relevant to the transplant population, such as adjustment of immunosuppressive regimens to limit progression of CKD and therapeutic considerations to minimize drug-drug interactions for management of highly prevalent comorbid conditions (eg, diabetes, hypertension, proteinuria) in transplant recipients. 11 Despite the breadth of recommendations for CKD management and high prevalence of CKD in LTRs, data on transplant provider management of CKD among LTRs are limited. Therefore, this study sought to assess rates of nephrology comanagement for CKD and utilization of guideline-recommended medical therapy among LTRs with CKD and whether these metrics are associated with a reduction in CV events.

Study Design
A longitudinal inception cohort study was conducted at a large, urban, tertiary care network in the United States. The institutional review board of Northwestern University approved the study.

Study Population
Patients who underwent LT between January 1, 2010, and December 31, 2016, were included in the study. We excluded patients who died within the first 6 mo after LT in an attempt to study patients with stable immunosuppression and graft function.

Data Source and Collection
Eligible LTRs were identified using International Classification of Diseases 9th or 10th Revision (ICD-9/10) codes and clinical information was obtained from the Northwestern Medicine Enterprise Data Warehouse, which contains comprehensive demographic, clinical, diagnostic, procedural, and administrative data for 7.5 million unique patients from all Northwestern Medicine sites. In addition, manual chart review was used for data that are not easily captured in an electronic health record (EHR), such as clinical reasoning for not adhering to a guideline recommendation (eg, documentation of why it would be inappropriate given the unique clinical scenario). Vital status was obtained from the Organ Procurement and Transplantation Network database, which is linked to the US Social Security Death Index. Data were linked to clinical data of each LTR based on a previously published methodology. 5,12

CKD and Covariate Definitions
Renal function was assessed by serum creatinine and corresponding eGFR calculated by the Modification of Diet in Renal Disease 4-variable equation. 13 CKD was identified by ICD-9/10 code or by using eGFR on at least 2 separate outpatient visits separated by ≥90 d. 14,15 CKD was defined as eGFR <60 mL/min/1.73 m 2 . CKD plus those at risk was defined as eGFR <90 mL/min/1.73 m 2 . At risk for CKD was defined as eGFR 60-89 mL/min/1.73 m 2 . Renal replacement status pre-and post-LT was defined as at least 2 encounters for intermittent hemodialysis or continuous renal replacement therapy before or after LT. Patients undergoing renal replacement were assigned an eGFR of 0 mL/min/1.73 m 2 , given the serum creatinine did not reflect actual renal function. Hypertension was identified by ICD-9/10 code, order of blood pressure (BP)-lowering medication, or systolic BP ≥140 or diastolic BP ≥90 mm Hg on at least 2 separate outpatient visits, consistent with clinical practice guidelines during the study period. Diabetes was identified by ICD-9/10 code, hemoglobin A1C ≥6.5%, random blood glucose >200 mg/ dL, or use of glucose-lowering medication, in the setting of prednisone daily dose ≤10 mg. Atherosclerotic cardiovascular disease (ASCVD) was identified by ICD-9/10 code for acute coronary syndrome, myocardial infarction, stable or unstable angina, coronary or other arterial revascularization, stroke, or transient ischemic attack. Obesity was defined as body mass index ≥30 kg/m 2 or ICD-9/10 code. Hyperlipidemia was identified by ICD-9/10 code, treatment with lipid-lowering medications, or total cholesterol ≥200 mg/dL. The standard model for end-stage liver disease (MELD) score was used because this was the allocation system in use during the study period. The immunosuppression and clinical visit protocol at our institution has been described previously. 16

Assessment of CKD Management
We used manual chart review to assess rates of nephrology comanagement among LTRs. Nephrology comanagement of CKD was defined as any referral to a nephrology specialist placed by a transplant provider during the study period or if a patient was already under the care of a nephrologist during the study period. We also assessed guideline-recommended medication use and average annual BP, using both manual and electronic chart review as described previously. 16

Exposure and Outcome Measures
The primary exposure variable was comanagement of CKD plus at-risk CKD by a nephrology specialist. The primary outcome variable was a CV event, defined as death from a CV cause or hospitalization for myocardial infarction/revascularization, cardiac arrest, heart failure, atrial fibrillation, thromboembolism, or stroke. In secondary analysis, we examined differences in rates of comanagement, process measures for CVD care, and associations with CV events among the subgroups with CKD (eGFR, <60) and those at risk for CKD (eGFR, 60-89).

Statistical Analysis
A t test, chi-square test, or Fisher exact test was used to examine group differences by nephrology comanagement status for continuous or categorical variables, as appropriate. Study participants without a serum creatinine recorded post-LT were excluded from this analysis (n = 30). Cox proportional hazard models were used to estimate major CV events from time of LT between LTRs with CKD who were comanaged by a nephrologist and those who were not. Time of LT was taken as time zero as the majority of CKD diagnoses occurred within the first year of LT and a substantial proportion (20%) of patients were already under the care of a nephrologist at the time of LT. The proportional hazard assumption was met and residuals were normally distributed. The model was adjusted a priori for sex, race, age at transplant and time-varying diabetes, ASCVD, use of BP-lowering medication, and CKD stage. SAS software version 9.4 (SAS Institute, Cary, NC) was used to complete all analyses. All P values are 2-sided and a P < 0.05 was considered to indicate statistical significance.
Cohort characteristics of LTRs with CKD or at risk for CKD stratified by nephrology comanagement status are shown in Table 1. The average age at time of LT was 57 ± 11 y, 60.7% of LTRs were men, 64.1% of LTRs identified as non-Hispanic White, and 17.9% were of Hispanic ethnicity. The most common indication for LT was hepatitis C (32.4%), followed by alcohol (22.1%), autoimmune (13.1%), and nonalcoholic steatohepatitis (12.9%). The mean MELD score at transplant was 24. LTRs with CKD or at risk for CKD who were comanaged with nephrology had higher rates of comorbid diabetes (43% versus 30%, P = 0.006) and higher mean MELD at time of LT (27.1 versus 21.7, P < 0.0001) compared with those that were not. There was no difference in prevalence of hypertension (59% with nephrology comanagement versus 53% without, P = 0.16), mean systolic BP level (133.9 mm Hg with comanagement versus 131.9 mm Hg without, P = 0.051), or BP medication usage (59% with comanagement versus 61% without, P = 0.78) by nephrology comanagement status.

Process of Care Measures
The rates of nephrology comanagement among those with any stage or at risk for CKD decreased yearly post-LT from 35% in year 1 to 28% in year 6. Rates of nephrology comanagement among those with CKD ranged from 30% to 42% annually post-LT. Table 2 demonstrates adherence to clinical practice guidelines yearly after transplant among LTRs who were comanaged by nephrology and those who were not. Among all LTRs with or at risk for CKD, only 5%-10% had a serum creatinine and a urine albumin:creatinine ratio obtained annually as recommended by clinical practice guidelines. 10 Less than 5% of LTRs with or at risk for CKD were offered a low-salt diet at least once yearly. Angiotensin-converting enzyme inhibitor (ACEi) or angiotensin II receptor blocker (ARB) therapy was offered to 8%-24% of LTRs with CKD and 9%-33% in those at risk for CKD. Among LTRs with or at risk for CKD and diabetes, ACEi/ARB therapy was offered to 11%-39% annually. However, in stratified analysis, LTRs who were comanaged by a nephrologist and who were at risk or had CKD and had diabetes were offered ACEi or ARB therapy at a higher frequency than those that were not comanaged by a nephrologist (12%-53% versus 9%-29% annually, respectively). There was a statistically significant difference between these annual adherence rates for 3 of the 6 post-LT y (year 2 P = 0.03, year 4 P = 0.01, year 6 P = 0.03). Among LTRs with or at risk for CKD and with hypertension, most had uncontrolled hypertension with only 3%-14% achieving guideline-directed average BP goal of <130/80 mm Hg annually post-LT (Table 2). Tables S1A and S1B (SDC, http://links.lww.com/TXD/A366) show yearly adherence rates to clinical practice guidelines in the subgroups of those with CKD only and the at-risk group only stratified by nephrology comanagement status.

CV Events
During a median follow-up of 3.5 y from transplant, 14.3% of LTRs experienced a CV event. The incidence rate for CV events among LTRs with or at risk for CKD was 40.1 events per 1000 person-y of follow-up time. The unadjusted CV event incidence rate was higher among LTRs with CKD (n = 66, 47.7 events per 1000 person-y of follow-up time) compared with those at risk for CKD (n = 11, 20.5 events per 1000 person-y of follow-up time). In unadjusted analyses, comanagement by a nephrologist among LTRs with or at risk for CKD was not statistically associated with major CV events (hazard ratio, 0.67; 95% confidence interval [CI], 0.38-1.17). However, when the model was adjusted for potential confounders including age at LT, sex, race and time-varying diabetes, ASCVD, and CKD stage, nephrology comanagement was associated with a 42% lower incidence of major CV events (adjusted hazard ratio [aHR], 0.58; 95% CI, 0.42-0.82). When additionally adjusted for time-varying use of BP-lowering medications, the association was attenuated but remained significant (aHR, 0.57; 95% CI, 0.33-0.99). When the analysis was restricted to only LTRs with CKD (eGFR, <60) or only those at risk (eGFR, 60-89), nephrologist comanagement was not associated with lower major CV events (aHR, 1.05; 95% CI, 0.66-1.68 and aHR, 0.60; 95% CI, 0.27-1.34, respectively). In secondary analysis, offering of ACEi or ARB therapy among LTRs with or at risk for CKD was not associated with lower overall mortality (aHR, 1.34; 95% CI, 0.89-2.01) or major CV events (aHR, 1.31; 95% CI, 0.92-1.86). Similar findings were found among LTRs with CKD (aHR, 1.14; 95% CI, 0.69-1.90 for overall mortality and aHR, 1.03; 95% CI, 0.69-1.53 for major CV events). Data expressed as % unless otherwise noted. Hypertension defined by the ICD-9/10 diagnosis codes, use of BP-lowering medication, systolic blood pressure ≥140 mm Hg, or diastolic BP ≥90 mm Hg on at least 2 separate visit dates. Hyperlipidemia defined by ICD-9/10 codes, total cholesterol ≥200 mg/dL, or use of lipid-lowering medication. Diabetes defined by ICD-9/10 codes, A1c ≥6.5%, or use of glucose-lowering medication. ASCVD defined by ICD-9/10 code for acute coronary syndrome, myocardial infarction, stable or unstable angina, coronary or other arterial revascularization, stroke, or transient ischemic attack.

DISCUSSION
We observed a significant association between nephrology comanagement of persons with or at risk for CKD and fewer major CV events among LTRs. However, rates of nephrology referral in this population were low. In addition, rates of adherence to guideline-directed quality measures associated with CV risk reduction were almost uniformly low. These findings suggest the hypothesis that transplant provider practices need to be critically evaluated and redesigned to help facilitate comanagement of LTRs with CKD with nephrologists, which could improve adherence to CKD clinical practice guidelines and subsequent outcomes.
Current clinical practice guidelines for the general CKD population recommend nephrology referral for several indications including GFR <30 mL/min/1.73 m 2 , albuminuria, acute kidney injury, or progression of CKD. 10 ACEi or ARB therapy is recommended for patients with CKD and evidence of proteinuria (albumin excretion, >300 mg/24 h) or diabetic nephropathy, which has been shown to improve proteinuria and delay CKD progression. 17,18 Although studies on ACEi/ ARB therapy have failed to show a reduction in major CV events when used for CKD, 17,19 these classes of medications have been shown to improve outcomes when used for several cardiac indications including post-myocardial infarction 20 and heart failure with reduced ejection fraction. 21,22 Despite these guidelines, only 8.7% of patients with stage 3, 39% with stage 4, and 32% with stage 5 CKD were seeing a nephrology specialist for comanagement of CKD in the Veterans Affairs Health System. 23 This is similar to the rates we found for nephrology comanagement of CKD among LTRs, with overall low rates of nephrology comanagement that are slightly higher for those with true CKD compared with those with CKD plus those at increased risk (30%-42% versus 28%-35%). Recent work by our group identified several barriers to providing multidisciplinary CV disease prevention care to LTRs, including (1) lack of awareness of CV disease risk after LT, (2) lack of confidence in an ability to provide proper CV disease care to LTRs, (3) reluctance to provide CVD care without transplant provider review, and (4) complexity of communication with the multidisciplinary LTRs care team about CVD care. 24 These identified barriers provide potential targets for quality improvement initiatives to improve comanagement of prevalent CV disease risk factors, such as CKD, among LTRs.
Although comanagement of CKD and its association with clinical outcomes has not previously been studied in LTRs, it has been associated with better outcomes in the general CKD population. A systematic review of 27 longitudinal cohort studies found that early nephrology referral for management of CKD was associated with lower hospitalization and mortality rates. 9 The results of the present study extend these findings into the LT population and show that nephrology comanagement was borderline significantly associated with a reduction in CV events, after adjusting for age, sex, race, Hypertension defined by ICD-9/10 diagnosis codes or use of BP-lowering medication or systolic BP ≥140 mm Hg or diastolic BP ≥90 mm Hg on at least 2 separate visit dates. Diabetes defined by ICD-9/10 codes or A1c ≥6.5% or use of glucose-lowering medication. a P value < 0.05. Fisher exact test or χ 2 test used where appropriate. ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; BP, blood pressure; CKD, chronic kidney disease; Cr, creatinine; DM, diabetes; eGFR, estimated glomerular filtration rate; HTN, hypertension; ICD-9/10, International Classification of Diseases, 9th and 10th Revisions; Ur, urine; + comanagement, nephrology referral placed or already established with nephrologist; − comanagement, nephrology referral not placed and not already established with nephrologist.