Association of Pre-Operative Albuminuria with Post-Operative Outcomes after Coronary Artery Bypass Grafting

The effect on post-operative outcomes after coronary artery bypass graft(CABG) surgery is not clear. Among 17,812 patients who underwent CABG during October 1,2006-September 28,2012 in any Department of US Veterans Affairs(VA) hospital, we identified 5,968 with available preoperative urine albumin-creatinine ratio(UACR) measurements. We examined the association of UACR<30, 30–299 and >=300 mg/g with 30/90/180/365-day and overall all-cause mortality, and hospitalization length >10 days, and with acute kidney injury(AKI). Mean ± SD baseline age and eGFR were 66 ± 8 years and 77 ± 19 ml/min/1.73 m2, respectively. 788 patients(13.2%) died during a median follow-up of 3.2 years, and 26.8% patients developed AKI(23.1%-Stage 1; 2.9%-Stage 2; 0.8%-Stage 3) within 30 days of CABG. The median lengths of stay were 8 days(IQR: 6–13 days), 10 days(IQR: 7–14 days) and 12 days(IQR: 8–19 days) for groups with UACR < 30 mg/g, 30–299 mg/g and ≥300 mg/g, respectively. Higher UACR conferred 72 to 85% higher 90-, 180-, and 365-day mortality compared to UACR<30 mg/g (odds ratio and 95% confidence interval for UACR≥300 vs. <30 mg/g: 1.72(1.01–2.95); 1.85(1.14–3.01); 1.74(1.15–2.61), respectively). Higher UACR was also associated with significantly longer hospitalizations and higher incidence of all stages of AKI. Higher UACR is associated with significantly higher odds of mortality, longer post-CABG hospitalization, and higher AKI incidence.

Scientific RepoRts | 5:16458 | DOI: 10.1038/srep16458 have emphasized the importance of pre-operative eGFR [9][10][11] and CKD with dipstick proteinuria 12 in predicting the risk of post-CABG AKI. However, these studies are limited by relative short-term follow up. It is also unknown if the same relationship exists in patients with eGFR ≥ 60 ml/min/1.73 m 2 and if it would be applicable to a larger population in a different geographic area.
We hence examined the association of pre-operative albuminuria with short-and long-term mortality and length of hospital stay, and also with post-operative AKI in a large cohort of US veterans undergoing CABG. We hypothesized that the degree of albuminuria is associated with proportionally higher risk of the studied outcomes.

Results
Baseline characteristics. The mean ± SD age of the cohort at baseline was 66 ± 8 years, 85% and 10% of patients were white and black, respectively, 88% of the patients were diabetic and the mean baseline eGFR was 77 ± 19 ml/min/1.73 m 2 . Baseline characteristics of patients categorized by UACR status are shown in Table 1. The level of eGFR, and the prevalence of diabetes, CHF, stroke, peripheral arterial disease were progressively higher in patients with higher UACR. Use of ACEI or ARB pre-operatively or during in-hospital stay was higher in patients with higher UACR. Compared to the analytic cohort (n = 5,968), patients excluded because of missing UACR (n = 11,844) had lower BMI, and a lower proportion of diabetes and hypertension at the time of study entry (results not shown).

Discussion
We describe an independent association between preoperative albuminuria and 90/180/365 and overall mortality, length of hospitalization and AKI following CABG in this cohort with baseline eGFR ≥ 60 ml/ min/1.73 m 2 . These associations were independent of eGFR level and comorbid conditions. The various outcomes showed a graded association with increasing severity of albuminuria.
Post-operative AKI after CABG is common and it is more frequently developing if the patient is operated more than 24 hours after cardiac catheterization 13 . Similar to our findings, in cohorts of CKD patients preoperative increased levels of proteinuria were associated with higher risk of mortality in patients who developed AKI after cardiac surgery 8,14 . In our study, the 90-, 180-and 365-day mortality was higher in patients with severe preoperative albuminuria. Similar to the study by Huang et al. 12 , we did not find an association between albuminuria and 30-day mortality. This could be explained by the closer attention devoted to patients in the immediate post-operative period, or by the longer time needed for pathophysiologic effects linked to higher albuminuria to manifest clinically following CABG. Alternatively, the lower number of events during such a short time period may result in decreased power to detect significant differences, in spite of a nominally higher risk of 30-day mortality. A lower power may also explain why we did not detect statistically significant associations between moderately higher albuminuria (UACR 30-299 mg/g) and mortality risk up to 365-days after CABG, even though the associations were nominally elevated, and overall mortality was significantly higher in this subgroup when considering the higher number of total deaths occurring throughout the entire follow-up period of our cohort.
Prior studies of patients with advanced CKD reported an association between proteinuria and the risk of postoperative AKI 5,8,12,15,16 . Our study shows that even in patients with eGFR ≥ 60 ml/min/1.73 m 2 there is a higher risk of postoperative AKI, and a higher risk for more severe AKI associated with albuminuria  Table 1. Baseline characteristics of patients. Abbreviations: UACR-urine albumin creatinine ratio; SDstandard deviation; BMI-body mass index; SBP-systolic blood pressure; DBP-diastolic blood pressure; DM-diabetes mellitus; CHF-congestive heart failure; PAD-peripheral arterial disease; HIV-human immunodeficiency virus; ACEI-ACE inhibitor; ARB-angiotensin receptor antagonist.
after CABG surgery. Moreover, the magnitude of this risk increases with increasing degree of albuminuria and the presence of severe albuminuria (UACR> = 300 mg/g) confers the highest risk of AKI in the post-operative period. Hence preoperative albuminuria should be an important variable in risk prediction models for postoperative AKI after CABG and it should be included to the future prediction score for AKI. A score post-operative atrial fibrillation was recently validated 17 , however no score with albuminuria was developed for AKI. It is important to identify such inexpensive preoperative markers of AKI so the utility of preventive measures aimed at correcting such risk factors during the perioperative period could be tested in randomized controlled clinical trials. The practical relevance of post-operative AKI is underscored by several large multicenter studies 14,18,19 indicating that post-operative AKI is associated with increased risk of dialysis-requiring kidney failure, mortality, and increased hospital costs 20,21 .
A previous study indicated that the length of hospital and intensive care unit stay after cardiac surgery is increased in patients with preoperative proteinuria 4 . However it was not clear if including patients with advanced CKD influenced this association. Our study, which included patients with eGFR ≥ 60 ml/ min/1.73 m 2 , showed that the length of hospital stay after CABG in patients with albuminuria was significantly higher. It is unknown what proportion of this increased length of stay can be attributed to the development of AKI, which was also increasingly seen in patients with higher preoperative proteinuria. Irrespective of the exact mechanisms involved, the increased duration of hospitalization could directly contribute to higher costs, and could expose patients to more hospitalization-associated adverse events such as nosocomial infections.
Limitations of the study. This being an observational study, we can only report associations, and we cannot claim that UACR was indeed the cause of the worse clinical outcomes. Additionally, models could only be adjusted for confounders for which we had available data. Therefore, we cannot rule out residual confounding. The study population consisted of mostly male patients; hence, the results may not be generalizable to females. Because we did not have information about causes of death, we could not analyze associations with cause-specific mortality.

Conclusion
Presence of preoperative albuminuria is associated with poorer post-operative outcomes in patients undergoing CABG. Albuminuria should be included as an important prognostic factor in risk prediction models for measuring clinical outcomes after CABG. The strength of association and degree of external  validity of this risk associated with preoperative proteinuria on postoperative outcomes need to be further evaluated, and the effects of proteinuria-lowering strategies on post-CABG outcomes will need to be assessed in large prospective randomized controlled clinical trials.

Methods
Cohort Definition. Our study utilized data from a cohort study examining risk factors in patients with incident CKD (Racial and Cardiovascular Risk Anomalies in CKD (RCAV) study), as previously described 22,23   calculated according to the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) Equation 25 . The flowchart for patient selection for the study is displayed in Fig. 3 115-628 days)). Socio-demographic characteristics, comorbid conditions and laboratory characteristics were obtained as previously described [26][27][28][29] . Information about age, gender and race were obtained through the VA Corporate Data Warehouse (CDW) and from Medicare through the VA-Medicare data merge project. Baseline characteristics were defined on or immediately preceding the date of CABG. The final cohort was divided into 3 groups according to UACR level: < 30 mg/g, 30-299 mg/g and ≥ 300 mg/g.

Outcomes.
We defined three different outcomes: 1) 30-, 90-, 180-, and 365-day and overall mortality, 2) length of hospitalization, and 3) incidence and severity of AKI. Data on mortality was obtained from the VA Vital Status Files (VSF), which contain dates of death or last medical/administrative encounter from all sources in the VA system with sensitivity and specificity of 98.3% and 99.8%, respectively, as compared to the National Death Index as gold standard 30 . The length of hospitalization was analyzed as a continuous variable, and also by categorizing it as ≤ 10 and > 10 days. AKI was sub-classified according to the Acute Kidney Injury Network creatinine-based criteria (no AKI, grade 1, grade 2, and grade 3) 31,32 , without considering urine output.
Statistical Analysis. Descriptive analyses were performed and skewed variables were log-transformed.
Data were summarized using proportions, means ( ± standard deviation, SD) or medians (interquartile range, IQR) as appropriate. Categorical variables were analyzed with chi-square test and continuous  Table 3. Multivariable adjusted association of UACR level with length of hospital stay and AKI. * p < 0.001. † p < 0.01. ‡ p < 0.05, compared to UACR < 30 mg/g group. Abbreviations: UACR-urine albumin creatinine ratio; AKI-acute kidney injury; OR-odds ratio; CI-confidence interval; SD-standard deviation. variables were compared using Student's t-test or the Mann-Whitney U test, Kruskal-Wallis H test or ANOVA as appropriate. In all statistics, two-sided tests were used and the results were considered statistically significant if p was < 0.05. The association of UACR with 30/90/180/365-day mortality, with AKI and with hospitalization length > 10 days was assessed using unadjusted and multivariable adjusted logistic regression models. To assess the association between UACR and overall mortality, we performed time-to-event analyses using Cox proportional regression models. In these analyses, patients were followed until death or were censored at the date of last healthcare or administrative visit, or on July 26, 2013. We also performed unadjusted and multivariable adjusted linear regression analyses to examine the association of UACR with length of hospital stay. Adjusted length of stay in each UACR category was predicted from multivariable regression models using Stata post-estimation commands. Non-linear associations were assessed using fractional polynomials and restricted cubic splines.
Models were adjusted sequentially for the following confounders based on a priori considerations: model 1: age, gender, race/ethnicity; model 2: model 1 variables and baseline systolic and diastolic blood pressure, administration of angiotensin converting enzyme inhibitors (ACEI) or angiotensin receptor blockers (ARB) pre-operatively and during the hospitalization, and administration of statins pre-operatively and during the hospitalization; model 3: model 2 variables and baseline eGFR, serum cholesterol, presence of diabetes mellitus, presence of congestive heart failure and the Charlson Comorbidity Index.
The propensity score method was used in a sensitivity analysis to account for baseline differences arising from dissimilarities in clinical and demographic characteristics of patients with and without albuminuria 33 . Variables associated with UACR were identified using logistic regression and were used to calculate propensity scores. STATA's "psmatch2" command suite was used to generate propensity score-matched cohorts by a 1-to-1 nearest neighbor matching without replacement. Statistical analyses were performed using STATA MP Version 12 (STATA Corporation, College Station, TX). The study protocol was approved by the institutional review boards at the Memphis and Long Beach Veterans Affairs Medical Centers and at the University of Tennessee Health Sciences Center.