Cystatin C Identifies Chronic Kidney Disease Patients at Higher Risk for Complications

Study Subjects

We used data from the MESA and the CHS that together provide a wide range of age, kidney function, and race/ethnicity. The MESA is an NHLBI, National Institutes of Health-sponsored study to understand subclinical cardiovascular disease and its progression in a racially diverse cohort. Details on recruitment and design have been previously published.²⁴ MESA recruited 6814 men and women who were between 45 and 84 years old, who were free of cardiovascular disease, and who self-identified as white, African American, Hispanic, or Chinese American. Subjects were recruited from Baltimore City and Baltimore County, Maryland; Chicago, Illinois; Forsyth County, North Carolina; Los Angeles County, California; Northern Manhattan and the Bronx, New York; and St. Paul, Minnesota between July 2000 and August 2002. Participants were excluded from the study if they had physician diagnosed heart attack, angina, heart failure, stroke, transient ischemic attack, or atrial fibrillation; had undergone coronary artery bypass grafting, angioplasty, valve replacement, or pacemaker; or weighed >300 lbs.²⁴

The CHS is a community-based longitudinal study designed to understand risk factors for the development and progression of cardiovascular disease. CHS recruited community-dwelling adults who were 65 years of age or older using Medicare eligibility lists in four sites: Forsyth County, North Carolina; Sacramento County, California; Washington County, Maryland; and Pittsburgh, Pennsylvania. Participants were excluded if they were not expected to remain in the current community for 3 years or longer, were receiving treatment for cancer, or were unable to provide informed consent. The initial 5201 participants were enrolled from January 1989 to June 1990; an additional 687 black participants (with race self-reported) were recruited and enrolled by June 1993. The study details and design have been published previously.²⁵ The institutional review boards at all participating centers approved these studies, and all participants gave informed consent. For these analyses, we included participants from CHS and MESA who had a serum creatinine and a serum cystatin C measured at baseline in MESA (n = 6749) or at the initial visit for CHS (i.e. 1989–1990 for original CHS cohort or 1992–1993 for the CHS black participants, n = 5160).

Kidney Function Measurements

All of the assays were performed in frozen serum specimens that were stored at −70°C. Cystatin C was measured by means of a particle-enhanced immunonephelometric assay (N Latex Cystatin C; Dade Behring) with a nephelometer (BNII; Dade Behring). Serum cystatin C was calibrated to Cleveland Clinic using internal standards supplied by the manufacturer to both sites. Serum creatinine was measured by a colorimetric method (Ektachem 700; Eastman Kodak) in CHS. In MESA, serum creatinine was measured by rate reflectance spectrophotometry using thin film adaptation of the creatine amidinohydrolase method on the Vitros analyzer (Johnson & Johnson Clinical Diagnostics, Inc., Rochester, New York) at the Collaborative Studies Clinical Laboratory at Fairview-University Medical Center (Minneapolis, Minnesota). Serum creatinine was calibrated directly to Cleveland Clinic in MESA and indirectly in CHS.^3,26 We estimated the GFR with the use of the newly developed CKD-EPI creatinine equation⁸ and the CKD-EPI cystatin C equation without demographic coefficients: eGFRcys = 76.7 × cystatin C^−1.19.¹⁷ Both formulae were developed from the pooling of several cohorts with GFR measured from iothalamate clearance. Urine albumin and creatinine were not available at baseline in CHS but were measured at year 7 in CHS using nephelometry.

Covariates

Age, gender, race, income, education, past or present smoking, and alcohol use were ascertained by questionnaire at the baseline visit. Height and weight were measured with participants wearing light clothing and no shoes. Body mass index was calculated as weight in kilograms divided by height in meters squared. Fasting blood was collected and stored at −70°F until needed for the appropriate assays, including HDL cholesterol, triglycerides, glucose, and C-reactive protein. LDL cholesterol was calculated using the Friedewald equation. Hypertension was defined as the use of antihypertensive medications, a self-report of hypertension, or a BP of >140/90 mmHg at the baseline visit. Diabetes was defined as a self-report of diabetes, the use of insulin or oral hypoglycemic agents, or a fasting glucose of ≥126 mg/dl. Three BP measurements were obtained 5 minutes apart in the seated position. The mean of the second two measurements was used for analysis. For CHS only, prevalent cardiovascular disease was defined as having a history of coronary heart disease, heart failure, or stroke.

Ascertainment of Outcomes

Outcomes of interest included: (1) death from all causes; (2) incident cardiovascular event (CVD), defined as myocardial infarction, cardiac arrest, stroke, or cardiovascular death; (3) incident heart failure (for CHS only); and (4) incident kidney failure (for CHS only). Each of the four outcomes was considered separately.

In MESA, in addition to the follow-up visits, an interviewer contacted each participant or family member by telephone every 9 to 12 months. A trained interviewer inquired about interim hospital admissions, outpatient diagnoses of cardiovascular disease, and deaths. To verify self-reported diagnoses, MESA requested copies of medical records for participants who had been hospitalized or received an outpatient diagnosis of cardiovascular disease. Records were obtained on 98% of reported cardiovascular events associated with hospitalization. For participants who had died of cardiovascular causes outside the hospital, MESA conducted interviews with the next of kin and requested copies of death certificates. Two physicians who were members of the MESA mortality and morbidity review committee independently classified events and assigned incidence dates. If they disagreed, the full committee made the final classification. Ascertainment of stroke was on the basis of clinical symptoms and brain imaging. A more detailed description of the MESA follow-up methods is available at www.mesa-nhlbi.org, and the methods have been previously published.^27,28 Median follow-up time was 4.7 years.

In CHS, follow-up visits were conducted by telephone every 6 months and in person annually. All of the events were adjudicated by a CHS outcome-assessment committee. Deaths were identified by a review of obituaries, medical records, death certificates, and the Centers for Medicare and Medicaid Services health care-utilization database for hospitalizations and from household contacts; 100% complete follow-up for ascertainment of mortality status was achieved. Cases of CVD events were ascertained from hospital records that included clinical histories, elevated cardiac enzyme levels, electrocardiographic changes, and brain imaging studies. Incident heart failure was adjudicated on the basis of diagnosis from a physician and consideration of symptoms, signs, chest radiographic findings, and treatment of heart failure. More details on event adjudication have been previously published.¹⁴ Kidney failure was identified by linking the CHS cohort to the United States Renal Data System (USRDS) in 2005 (which includes data through March 31, 2003) to identify individuals who initiated dialysis or underwent kidney transplantation. In addition to linking the CHS cohort to the USRDS, a chart review was performed to determine which CHS participants met the criteria for initiation of dialysis or transplantation, elected not to undergo these therapies, died before receiving them, or died before being enrolled in the USRDS. Median follow-up time was 12.2 years.

Statistical Analyses

Participant characteristics were summarized by cohort. We first estimated GFR using equations on the basis of creatinine and by cystatin C separately for each individual. We then categorized individuals into four groups defined by presence or absence of eGFR <60 ml/min per 1.73 m² on the basis of cystatin C and creatinine: (1) those with eGFRcreat <60 ml/min per 1.73 m², and eGFRcys <60 ml/min per 1.73 m² herein described as decreased GFR both; (2) those with eGFRcreat ≥60 ml/min per 1.73 m² but eGFRcys <60 ml/min per 1.73 m², herein described as decreased GFRcys only; (3) those with eGFRcreat <60 ml/min per 1.73 m² but eGFRcys ≥60 ml/min per 1.73 m², herein described as decreased GFRcreat only; and (4) those with eGFRcreat ≥60 ml/min per 1.73 m², and eGFRcys ≥60 ml/min, herein described as GFR not decreased. We estimated the prevalence of each group in CHS and MESA separately.

For each of the above four groups, we estimated the incidence rates of death and cardiovascular disease in MESA and CHS, and the rates of heart failure and kidney failure in CHS only. Then, using Cox proportional hazard models, we determined their association with the risks for death, cardiovascular events, incident heart failure, and kidney failure in separate models. We adjusted for the above-mentioned covariates chosen a priori from the literature as potential confounders of the association of eGFR <60 ml/min per 1.73 m² with adverse outcomes.

We used data from MESA and CHS combined to estimate the prevalence of eGFRcys <60 ml/min per 1.73 m² among persons with eGFRcreat <60 or ≥60 ml/min per 1.73 m², overall and by decade of age. Using these proportions, we estimated the number needed to screen to detect additional cases of CKD, which was calculated as (1/(# with decreased cys only/# with eGFRcreat ≥60 ml/min per 1.73 m²). We also calculated the number needed to confirm CKD, which was calculated as (1/ (# with decreased GFR both /# eGFRcreat <60 ml/min per 1.73 m²). We stratified these analyses by ages.

We also estimated the net reclassification improvement for eGFRcys.²⁹ To this end, we constructed a risk prediction model for death including eGFRcreat <60 ml/min per 1.73 m² as the primary predictor and adjusting for age, gender, race, diabetes, hypertension, LDL and HDL cholesterol, smoking status, body mass index, and C-reactive protein. We classified persons into three categories of mortality risk in CHS (10 years risk of death <10%, 10 to 20%, or >20%) and MESA (5 years risk of death <5%, 5 to 10%, or >10%) separately. We replaced eGFRcreat with eGFRcys in the model and estimated the reclassification of persons into higher or lower risk strata. We calculated the net reclassification improvement as the sum of the proportions of correctly reclassified subjects with and without death.

Based upon the results of the multivariate analyses, we used CART to determine the likelihood of eGFRcys <60 ml/min per 1.73 m², on the basis of combinations of age, gender, race, and serum creatinine. CART is on the basis of recursive partitioning and is an empirical method that can generate decision trees with binary splits for best model fit.³⁰

In a secondary analysis, we compared the prevalences of albuminuria (defined as albumin to creatinine ratio >30 mg/g) and eGFRcys <60 ml/min per 1.73 m² among CHS participants with GFRcreat <60 ml/min per 1.73 m². We used data from the CHS year 7 exam, because urine albumin and creatinine were not available at baseline. We categorized the CHS participants with eGFRcreat <60 ml/min per 1.73 m² into four mutually exclusive groups: (1) eGFRcys ≥60 ml/min per 1.73 m² and no albuminuria; (2) eGFRcys ≥60 ml/min per 1.73 m² and albuminuria; (3) eGFRcys <60 ml/min per 1.73 m² and no albuminuria; and (4) eGFRcys <60 ml/min per 1.73 m² and albuminuria. We estimated the age adjusted mortality rates for each group and used Cox proportional hazards to determine adjusted associations with mortality risk for each group.

Finally, we performed a sensitivity analysis using the cystatin C eGFR estimating equation that includes demographic coefficients.¹⁷ We divided the MESA and CHS participants into the four mutually exclusive groups on the basis of eGFRcys and eGFRcr, and we replicated our primary analyses.

All of the analyses were performed using S-Plus (release 8.0; Insightful Inc, Seattle, WA) and SPSS statistical software (release 16.0.1; SPSS Inc, Chicago, IL). Two-tailed P values <0.05 were considered significant.