Creatinine versus cystatin C for renal function-based mortality prediction in an elderly cohort: The Northern Manhattan Study.

BACKGROUND: Estimated glomerular filtration rate (eGFR) is routinely utilized as a measure of renal function. While creatinine-based eGFR (eGFRcr) is widely used in clinical practice, the use of cystatin-C to estimate GFR (eGFRcys) has demonstrated superior risk prediction in various populations. Prior studies that derived eGFR formulas have infrequently included high proportions of elderly, African-Americans, and Hispanics.
OBJECTIVE: Our objective as to compare mortality risk prediction using eGFRcr and eGFRcys in an elderly, race/ethnically diverse population.
DESIGN: The Northern Manhattan Study (NOMAS) is a multiethnic prospective cohort of elderly stroke-free individuals consisting of a total of 3,298 participants recruited between 1993 and 2001, with a median follow-up of 18 years. PARTICIPANTS: We included all Northern Manhattan Study (NOMAS) participants with concurrent measured creatinine and cystatin-C. MAIN MEASURES: The eGFRcr was calculated using the CKD-EPI 2009 equation. eGFRcys was calculated using the CKD-EPI 2012 equations. The performance of each eGFR formula in predicting mortality risk was tested using receiver-operating characteristics, calibration and reclassification. Net reclassification improvement (NRI) was calculated based on the Reynolds 10 year risk score from adjusted Cox models with mortality as an outcome. The primary hypothesis was that eGFRcys would better predict mortality than eGFRcr.
RESULTS: Participants (n = 2988) had a mean age of 69±10.2 years and were predominantly Hispanic (53%), overweight (69%), and current or former smokers (53% combined). The mean eGFRcr (74.68±18.8 ml/min/1.73m2) was higher than eGFRcys (51.72±17.2 ml/min/1.73m2). During a mean of 13.0±5.6 years of follow-up, 53% of the cohort had died. The AUC of eGFRcys (0.73) was greater than for eGFRcr (0.67, p for difference<0.0001). The proportions of correct reclassification (NRI) based on 10 year mortality for the model with eGFRcys compared to the model with eGFRcr were 4.2% (p = 0.002).
CONCLUSIONS: In an elderly, race/ethnically diverse cohort low eGFR is associated with risk of all-cause mortality. Estimated GFR based on serum cystatin-C, in comparison to serum creatinine, was a better predictor of all-cause mortality.

Introduction

The prevalence of chronic kidney disease (CKD) increases dramatically among the elderly [1, 2] and has been identified by several investigators as a risk factor for cardiovascular disease (CVD) related outcomes including mortality [3, 4], heart failure, myocardial infarction [5], stroke [6], and cognition [7-9]; it is furthermore linked to frailty [10, 11]. The impact of CKD on CVD outcomes is independent of their shared risk factors, such as hypertension and diabetes, and in excess of other known risk factors including prevalent CVD [5, 12]. In addition, the increased mortality observed in diabetic patients is predominantly accounted for by the presence of CKD [12]. Furthermore, CKD has a disproportionate burden among those with lower socio-economic status, blacks and Hispanics [13], and may partly explain the increased medication adverse events seen in elderly blacks and Hispanics [14, 15]. Despite these well-documented consequences of CKD, there is a paucity of data in elderly diverse cohorts on the prevalence of CKD as well as the impact of CVD. Furthermore, it is not well known if in elderly diverse populations estimated glomerular filtration (eGFR) calculations using either serum creatinine or cystatin-C can adequately predict CVD and mortality. Previously we and others have shown that eGFR equations using creatinine or cystatin-C can provide significantly divergent estimates of the prevalence of CKD [16]. The goals of this study were to examine 1) the association of CKD using eGFR from creatinine (eGFRcr) or eGFR from cystatin-C (eGFRcys)with CVD and mortality in an elderly race/ethnically diverse cohort, and 2) performance of eGFRcr and eGFRcys in predicting mortality risk. We hypothesized that a eGFRcys would predict risk of mortality more accurately compared to eGFRcr.

Methods

Recruitment of the cohort

The recruitment and assessment of the Northern Manhattan Study (NOMAS) cohort has been described in previous publications [17]. Briefly, eligible participants were: 1) stroke free; 2) resident of at least 3 months duration of Northern Manhattan as defined by zip-codes 10031, 10032, 10033, 10034, & 10040; 3) randomly derived from a household with a telephone; 4) age 40 years or older (changed to age 55 or older in 1998) at the time of first in-person assessment. Participants were recruited between 1993–2001and followed longitudinally to present date. All participants gave informed consent to participate in the study. Race-ethnicity was determined by self-identification and standardized questions were used regarding hypertension, diabetes, cigarette smoking, alcohol intake and cardiac comorbidities. Blood pressure was measured twice, before and after each examination, and averaged. Hypertension was defined as a blood pressure ≥140/90 mmHg, the patient’s self-report of hypertension, or use of anti-hypertensive medications. Diabetes mellitus was defined by the patient’s self-report of a history of diabetes, use of insulin or oral anti-diabetic medication, or fasting glucose ≥126 mg/dl. Hypercholesterolemia was defined as having a total cholesterol level of greater than 200 mg/dl, use of cholesterol lowering medications, or self-reported history of hypercholesterolemia. The study was approved by the institutional review boards of Columbia University Irving Medical Center and the University of Miami.

Measurement of creatinine and cystatin-C

Blood samples were obtained during baseline enrollment from 1993–2001. All laboratory testing was performed at Columbia University Medical Center or at the University of Miami. Serum creatinine (mg/dL) was measured using Olympus instrumentation with a Jaffe-based method. Although the initial creatinine concentrations were measured prior to the isotope dilution mass spectroscopy (IDMS) standardization for estimated GFR, creatinine was re-measured in 100 samples stored at -80°C using an IDMS-traceable method for creatinine measurement in order to develop a correction factor similar to what had been done successfully by other cohorts [18] [19]. The mean difference between standardized and non-standardized creatinine was -0.056 ± 0.079 mg/dL. In the absence of a meaningful difference, a calibration factor was not applied prior to using the creatinine for GFR estimation using the CKD-EPI 2009 equation [20]. However, a sensitivity analysis was performed by repeating the primary analysis using creatinine values after calibration factor application. Cystatin-C (mg/L) was measured on samples (84% plasma, 14% serum, 2% unspecified) stored at -80°C using Roche Diagnostics Cystatin Reagents on a Roche analyzer, standardized against ERM-DA471/IFCC reference material (intra-assay coefficient of variation (CV) of 2.8% and an inter-assay CV of 4.1%; reference range 0.5–1.3 mg/L). Cystatin-based GFR estimation used the CKD-EPI 2012 equation [21].

Statistical analysis

The primary outcome of interest was all-cause mortality, with secondary outcomes of vascular mortality, non-vascular mortality, stroke, MI, and a combined vascular outcome (stroke, MI, vascular death). The association of eGFR, defined as < 60 ml/min/1.73m2 and per 10 ml/min/1.73m2, with the outcomes in this study was examined using Cox proportional hazard models to calculate hazards ratios (HR) and 95% confidence intervals (CI). The models were first calculated unadjusted and then followed by adjusting for cardiovascular disease risk factors (age, sex, race-ethnicity, education, Medicaid/no insurance, diabetes, hypertension, body-mass index, tobacco use, hypercholesterolemia, and heart disease). In order to examine the performance in mortality risk prediction for eGFRcr and eGFRcys, we constructed two models: 1) a model with continuous eGFRcr as a main predictor, and 2) a model with continuous eGFRcys as a main predictor. We compared receiver-operator characteristic (ROC) curves by treating mortality as a binary outcome, and calculated area under the curve (AUC). To account for censoring, we additionally examined whether AUC changed over time. We also compared the estimated Net Reclassification Improvement (NRI) based on Reynold’s 5-year and 10-year mortality risk scores given the high proportion of women in our cohort [22, 23]. Reynold’s mortality risk scores were calculated using adjusted Cox proportional hazard models with mortality as an outcome, and then the calculated predicted mortality risk probabilities were categorized as <5%, 5–10%, 10–20% and >20% in order to examine NRI. We further examined the modification effect of NRI by age <70 vs > = 70, sex and race-ethnicity.

Results

There were 2988 participants with both serum creatinine and cystatin-C available. The mean age was 69±10.2 years and participants were predominantly Hispanic (53%) or black (24%), overweight (69%), and current or former smokers (53% combined). The mean eGFRcr (74.68±18.8 ml/min/1.73m2) was higher than eGFRcys (51.72±17.2 ml/min/1.73m2); there was a greater difference in GFR estimations at the upper rather than lower ranges (Figs 1 and 2).

Fig 1

Dot plot of estimated glomerular filtration rate using serum creatinine and cystatin-C.

Fig 2

Bland-Altman plot of estimated glomerular filtration rate using serum creatinine and cystatin-C.

Baseline characteristics are summarized in Table 1. Over a mean of 13 years there were 350 strokes, 208 myocardial infarctions, 475 vascular deaths, 810 non-vascular deaths, and 1611 all-cause deaths (n = 326 unclassified deaths).

Table 1

Baseline demographics of the Northern Manhattan Study.

	Mean (standard deviation) or No. (proportion as %)
Age, years, mean (standard deviation)	69 (10.2)
Male	1101 (37%)
Non-Hispanic black	725 (24%)
Non-Hispanic white	619 (21%)
Hispanic	1577 (53%)
Education (completed high school)	1377(46%)
Medicaid/no insurance	1287 (43%)
Diabetes	634 (21%)
Hypertension	2196 (74%)
Body-mass index, mean (std)	27.8 (5.5)
Active tobacco	498 (17%)
Prior tobacco use	1084 (36%)
Hypercholesterolemia	1893 (63%)
Heart disease	704 (24%)
Serum creatinine, mg/dL	0.96 (0.4)
Serum cystatin C, mg/L	1.4 (0.6)
eGFRcr ml/min/1.73m2	74.68 (18.8)
eGFRcys ml/min/1.73m2	51.72 (17.2)

Association of eGFR with outcomes

In unadjusted models we found that eGFRcr<60 ml/min/1.73m2 and eGFRcys<60 ml/min/1.73m2 were both associated with an increased risk of vascular and non-vascular mortality, as well as the combined vascular endpoint of stroke/MI/vascular death (Table 2). In multi-variable models the associations were somewhat attenuated but remained significant for all-cause mortality for both eGFRcr<60 ml/min/1.73m2 (adjusted HR 1.24, 95%CI 1.11–1.39) and eGFRcys<60 ml/min/1.73m2 (adjusted HR 1.41, 95% CI 1.22–1.63). eGFRcys<60 ml/min/1.73m2 was associated with vascular and non-vascular mortality; eGFRcr was only associated with non-vascular mortality. Both estimates of GFR< 60 were associated with the combined vascular end-point. The eGFRcr< 60 ml/min/1.73m2 (adjusted HR 1.50, 95% CI 1.09–2.06), but not eGFRcys< 60 ml/min/1.73m2 (adjusted HR 1.38, 95% CI 0.96–2.00), was associated with MI. Neither of the estimates of eGFR< 60 ml/min/1.73m2 was associated with risk of stroke in adjusted models. The results examining eGFR per 10 ml/min/1.73m2 increments were similar to the categorical definitions for eGFR except for eGFRcr no longer being associated with the combined end-point, and eGFRcys being associated with the risk of MI (Table 2).

Table 2

Associations of estimated glomerular filtration using serum creatinine and cystatin-C with mortality and vascular outcomes in the Northern Manhattan Study.

	eGFRcr < 60 ml/min/1.73m2 (unadjusted hazards ratio, 95% confidence interval)	eGFRcr < 60 ml/min/1.73m2 (adjusted hazards ratio, 95% confidence interval)*	eGFRcys < 60 ml/min/1.73m2 (unadjusted hazards ratio, 95% confidence interval)	eGFRcys < 60 ml/min/1.73m2 (adjusted hazards ratio, 95% confidence interval)*	eGFRcr (per 10ml/min/1.73m2 increase) (unadjusted hazards ratio, 95% confidence interval)	eGFRcys (per 10ml/min/1.73m2 increase) (unadjusted hazards ratio, 95% confidence interval)	eGFRcr (per 10ml/min/1.73m2 increase) *(adjusted hazards ratio, 95% confidence interval)	eGFRcys (per 10ml/min/1.73m2 increase) *(adjusted hazards ratio, 95% confidence interval)
All-cause mortality	2.21, 1.99–2.46	1.24, 1.11–1.39	2.74, 2.41–3.12	1.41, 1.22–1.63	0.79, 0.76–0.81	0.64, 0.62–0.67	0.94, 0.91–0.97	0.80, 0.77–0.83
Vascular Death	2.17, 1.78–2.64	1.16, 0.94–1.44	3.00, 2.38–3.79	1.45, 1.11–1.88	0.80, 0.76–0.84	0.64, 0.60–0.68	0.97, 0.92–1.03	0.82, 0.76–0.88
Non-vascular death	2.01, 1.73–2.34	1.20, 1.02–1.41	2.41, 2.02–2.87	1.35, 1.11–1.65	0.80, 0.77–0.83	0.66, 0.63–0.69	0.94, 0.89–0.98	0.80, 0.75–0.84
Stroke	1.76, 1.39–2.22	1.17, 0.90–1.51	1.51, 1.19–1.91	0.98, 0.74–1.28	0.85, 0.80–0.90	0.83, 0.78–0.89	0.95, 0.89–1.02	0.99, 0.91–1.07
Myocardial infarction	2.24, 1.67–3.02	1.50, 1.09–2.06	2.03, 1.47–2.81	1.38, 0.96–2.00	0.79, 0.74–0.86	0.73, 0.67–0.80	0.89, 0.81–0.96	0.83, 0.74–0.93
Combined vascular endpoint**	2.08, 1.77–2.45	1.22, 1.02–1.46	2.28, 1.92–2.72	1.28, 1.05–1.57	0.82, 0.79–0.85	0.71, 0.67–0.74	0.97, 0.92–1.01	0.88, 0.83–0.93

*Adjusted for age, sex, race-ethnicity, education, Medicaid/no insurance, diabetes, hypertension, body-mass index, tobacco use, hypercholesterolemia, and heart disease

**Stroke, myocardial infarction, vascular death

Comparison in mortality risk predictions

In order to examine the performance in mortality risk prediction, we first compared the model fit using ROC curves (Fig 3).

Fig 3

Area under the curve (AUC) comparisons for model fit for all-cause mortality of glomerular filtration rate estimated by serum creatinine (eGFRcr) and cystatin-C (eGFRcys) at five and ten years of follow up.

We found that eGFRcys was associated with an improved model performance compared to eGFRcr (AUC 0.73 vs 0.67, p for difference< 0.0001) when mortality was treated as a binary outcome.

We further examined whether the AUC for models with each eGFR was changing over the years of follow-up, and found no significant change in AUC over time (Figs 4 and 5).

Fig 4

Comparison of area under the curve for all-cause mortality at 5 years for glomerular filtration rate estimated by serum creatinine (eGFRcr) and cystatin-C (eGFRcys).

Fig 5

Comparison of area under the curve for all-cause mortality at 10 years for glomerular filtration rate estimated by serum creatinine (eGFRcr) and cystatin-C (eGFRcys).

The proportion of correct reclassification by the model with eGFRcys compared to the model with eGFRcr was 4% based on Reynold’s 10-year risk (p = 0.002) and 11% on 5-year risk (p < .0001), respectively (Tables 3 and 4).

Table 3

Comparison of the 5 year estimated mortality risk between the model with eGFRcys and the model with eGFRcr.

5-year mortality risk based on Model with eGFRcr	5-year mortality risk based Model with eGFRcys					Reclassified
	< 5%	5%~10%	10%~20%	> = 20%	Total	n (%)
< 5%
Participants, n (%)	854 (92.3)	71(7.7)	0(0.0)	0(0.0)	925	71 (7.7)
actual event rate (%)	2.7	7.0	0.0	0.0
5%~10%
Participants, n (%)	122 (16.8)	522 (71.9)	82 (11.3)	0 (0.0)	726	204 (28.1)
actual event rate (%)	2.5	5.0	19.5	0.0
10%~20%
Participants, n (%)	1 (0.2)	93 (14.5)	482 (75.2)	65 (10.1)	641	159 (24.8)
actual event rate (%)	0.0	5.4	13.7	44.6
> = 20%
Participants, n (%)	0 (0.0)	1 (0.2)	63 (10.8)	521 (89.1)	585	64 (10.9)
actual event rate (%)	0.0	0.0	20.6	37.0

Table 4

Comparison of the 10 year estimated mortality risk between the model with eGFRcys and the model with eGFRcr.

10 year mortality risk based on Model with eGFRcr	10 year mortality risk based on Model with eGFRcys					Reclassified
	< 5%	5%~10%	10%~20%	> = 20%	Total	n
< 5%
Participants, n (%)	174 (87.0)	26 (13.0)	0 (0.0)	0 (0.0)	200	26 (13.0)
actual event rate (%)	2.9	0.0	0.0	0.0
5%~10%
Participants, n (%)	77 (15.5)	358 (72.0)	61 (12.3)	1 (0.2)	497	139 (28.0)
actual event rate (%)	5.2	6.1	16.4	0.0
10%~20%
Participants, n (%)	3 (0.4)	108 (15.4)	508 (72.6)	81 (11.6)	700	192 (27.4)
actual event rate (%)	0.0	13.9	13.0	21.0
> = 20%
Participants, n (%)	0 (0.0)	1 (0.1)	96 (6.5)	1374 (93.4)	1471	97 (6.6)
actual event rate (%)	0.0	0.0	16.7	47.3
					2868	454

When the interactions of NRI with age<70 vs. > = 70 were examined, there was a significant difference in NRI based on 5 year mortality risk; the proportion of correct reclassification by the model with eGFRcys compared to eGFRcr was greater among those with age <70 than age> = 70 (estimated NRI = 22% for age<70 group and 9% for age> = 70 group: p for difference = 0.047) We also found an interaction by sex such that the proportion of correct reclassification was higher in men than in women (estimated NRI for women 7%, men 17%, p for difference = 0.049) with eGFRcys compared to eGFRcr.

The AUC for models with each GFR supported the improved model fit in this age group (Fig 6).

Fig 6

Area under the curve (AUC) comparisons for model fit for all-cause mortality of glomerular filtration rate estimated by serum creatinine (eGFRcr) and cystatin-C (eGFRcys) stratified by age.

We, however, found no interactions of NRI based on 10 year mortality risk by age groups. Similarly there were no modification effects of NRI by sex or race-ethnicity, and the AUC for each GFR equation by race-ethnicity was similar (Figs 7–20).

Fig 7

Area under the curve (AUC) comparisons for model fit for all-cause mortality of glomerular filtration rate estimated by serum creatinine (eGFRcr) and cystatin-C (eGFRcys), age < 70 years at 5 years.

Fig 20

Area under the curve (AUC) comparisons for model fit for all-cause mortality of glomerular filtration rate estimated by serum creatinine (eGFRcr) and cystatin-C (eGFRcys), Hispanics at 10 years.

Discussion

In an elderly race/ethnically diverse cohort with a large burden of hypertension and diabetes we found that CKD defined by either serum creatinine or cystatin-C based eGFR was associated with an increased risk of mortality, especially vascular death. However, no significant associations with non-fatal CVD events such as stroke or myocardial infarction were found. Though both estimates of eGFR were associated with the risk of death, the estimated GFR using serum cystatin-C was better in predicting 5-year and 10-year mortality risk. Notably, eGFRcys outperformed eGFRcr in predicting 5-year mortality risk especially among those with age <70 years; this same age group was more likely to be black and Hispanic compared to white.

Our results, in an elderly multiethnic cohort, are in keeping with the established association of CKD with mortality, particularly with cystatin-C based estimates in the Cardiovascular Health Study for example [24]. The results related to eGFRcys are consistent with findings from other studies suggesting that eGFRcys may be a more accurate estimate of GFR than a serum creatinine-based formula [25], and extend those findings to an elderly multiethnic population where GFRcr may be confounded by loss of muscle mass which would attenuate the association. The inability to accurately estimate GFR disproportionately affects women, blacks and Hispanic elderly patients creating significant challenges for prognostication for outcomes, decline of renal function, and management (particularly for medication dosing) of these individuals. For example, in the NOMAS cohort, creatinine and cystatin based eGFR have resulted in dramatically different estimates of CKD prevalence (21.9% and 70.5% respectively) calling into question the precision of the eGFR equations. Similar divergent results have been reported by other albeit smaller cohorts [26-28]. Interestingly in our cohort the predictive ability of eGFR (regardless of serum measure) appeared higher in the younger participants and men who were most likely to be included in prior cohort that derived GFR estimation formulae. These results highlight the importance of improved accuracy in measurement of GFR in diverse populations will help better understand how CKD is associated with CVD mortality related disparities.

The most widely accepted equations to estimate GFR using serum creatinine in adults include the Modification of Diet in Renal Disease (MDRD) and the more recent CKD Epidemiology Collaboration (CKD-EPI) equation. The latter, which includes the same variables as the MDRD equation but with different coefficients, has been described as a more accurate estimate of GFR across the range of renal function, especially for eGFR > 60mL/min, and provides better risk stratification in the general population [29, 30]. However, creatinine generation is directly related to muscle mass, and creatinine based eGFR estimation is therefore impacted by age and other circumstances that result in a change in body composition such as sarcopenia [31], limb loss, as well as functional impairments [32]. The estimation of GFR using serum cystatin-C, an endogenous protease inhibitor produced by all nucleated cells and filtered freely by the kidneys, has been proposed as a potentially more accurate method of renal function assessment and a better prognostication marker, particularly in the elderly and diverse populations [24, 33–36]. Serum cystatin-C based GFR estimation has not yet been widely adopted in clinical practice and more recently has been recognized to also be affected by aging, raising questions about the interpretation of GFR estimates in the elderly [37].

A further significant concern regarding currently used eGFR formulas is their generalizability to populations with substantial proportions of elderly and Hispanics. For example the Cockcroft-Gault formula was initially derived in 1979 in a study of only white men [38]. The MDRD equation was developed in a cohort of 1628 participants with a mean age of 50 ± 13 years, 60% men, and 88% non-Hispanic white [39]. The CKD-EPI equation was developed in 5504 participants including only <5% of the sample over age 70, though race-ethnicity representation was slightly improved (32% black, 5% Hispanic) [20]. Additional formulas have been proposed using both serum creatinine and cystatin-C with even lower proportions of elderly blacks and Hispanics [25, 40]. Newer estimates, including the Chronic Renal Insufficiency Cohort GFR estimating equation, performed poorly among Hispanics, blacks, and elderly [41]. Similarly, the Berlin Initiative Study (BIS) using both serum creatinine and cystatin-C (mean age 78.5) in a European population did not perform well in accurately predicting renal function in other populations with higher proportions non-whites [28]. Unfortunately, there is a paucity of data on the most accurate GFR formula to use in diverse populations despite prior research documenting differences in serum creatinine and cystatin-C by age and race-ethnicity [42]. Population-based studies in the United States with large proportions of diverse participants have been limited to smaller sample sizes such as the 294 participants in MESA-Kidney [43].

The strengths of our study include a large multi-ethnic population and comprehensive follow up for death and CVD events over 10 years. Our study has several weaknesses that require discussion. In NOMAS, we did not measure GFR with an exogenous marker such as iohexol or iothalamate and as such cannot determine which serum marker provides the most accurate estimate of the measured GFR in absolute terms. Instead, we focused on predictive validity, which may have other clinical advantages beyond the mere estimation of renal function. NOMAS did not obtain repeated measures of serum creatinine or cystatin-C to document a decline in values over time, nor did we systematically determine whether participants progressed to end stage renal disease. This data would provide additional information on which marker better predicted prevalent higher stages of CKD. Cystatin-C levels can be affected by several medical conditions including thyroid dysfunction [44] and human immunodeficiency virus infection [45] which unfortunately we did not collect in NOMAS. Lastly, we did not collect urine protein measurements that would help identify CKD in patients with eGFR >60mL/min or improve the risk prediction models.

In conclusion, eGFRcys provided a better prognostication tool for the risk of mortality compared to eGFRcr. Further research is required in diverse populations, including elderly multi-ethnic populations, on accurately measuring GFR.

19 Aug 2019

PONE-D-19-20793

Creatinine versus Cystatin C for Renal Function-Based Mortality Prediction in an Elderly Cohort: the Northern Manhattan Study

PLOS ONE

Dear Dr Willey,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

We would appreciate receiving your revised manuscript by Oct 03 2019 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Tatsuo Shimosawa, M.D., Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

1. When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The manuscript by Willey et al. investigated the role of estimated GFR based on creatinine (eGFRcr) or cystatin-C (eGFRcys) for mortality risk prediction in an elderly, ethnically diverse cohort, and found that eGFRcys was superior in predicting the risk of all-cause mortality. The topic may be of scientific interest, and might give some impact on clinical practice. I feel, however, there are still some unclear and unconvincing points that should be clarified from a scientific viewpoint. My major concerns are as follows:

1. Although there was a slight but significant difference in a predicting power between the two measures, the reason is not clear. As the authors state in Discussion, the most important point must be whether eGFRcys is superior in accurately measuring GFR in their cohort. This point should be clarified at least in a small number of subjects in their cohort, using iohexol or inulin as an exogenous marker.

2. Also, the dot plot between eGFRcr (mean, 74.7) and eGFRcys (mean, 51.7) should be presented to know the correlation between the two. As the CKD stage progresses, does the discrepancy become larger?

3. It is also important and should be analyzed whether the predicting power of eGFRcys was influenced by the severity of CKD, i.e., CKD stage 1-2 (eGFR > 60), stage 3, or stage 4-5 (eGFR < 30). Mortality incidence should have been more often in advanced CKD stages. This point should be clearly presented.

4. In Figure 3, the ROC curve of eGFRcys seemed much better than eGFRcr among the subjects under 70. Why?

5. It is not clear what are critical conditions where eGFRcys is superior to eGFRcr in mortality risk prediction. Males? Younger age? Caucasians? Subgroup analysis should be performed in order to specify the factors which have impact in favor of eGFRcys for risk prediction.

Reviewer #2: The authors compared creatinine and cystatin C as factor of eGFR. I think that the aim of this study is very interesting to nephrologist, and felt old.

1. I believe that authors knew serum cystatin C concentration is influenced by many factors, hyper/hypo thyroid, HIV infection, and so on. I could not find in the manuscript about exclusion of these patient from cohort.

2. Main finding of this study may be “eGFRcys predicted all-cause mortality better than eGFRcre”. However, this fact is already reported indirectly as the authors cited in Ref 24, 33,34, and Astor BC et al 2011, Peralta CA et al 2011

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

5 Nov 2019

We appreciate the reviewer’s comments on our manuscript. We have included the comments by the reviewers below and have modified the manuscript as requested in the appropriate sections and included below are responses to the reviews.

Reviewer #1:

1. Although there was a slight but significant difference in a predicting power between the two measures, the reason is not clear. As the authors state in Discussion, the most important point must be whether eGFRcys is superior in accurately measuring GFR in their cohort. This point should be clarified at least in a small number of subjects in their cohort, using iohexol or inulin as an exogenous marker.

Response:

We agree with the reviewer that having a gold-standard measure of GFR would be ideal. Unfortunately the serum measures were collected at the time of initial enrollment in the Northern Manhattan Study between 1993 to 2001 such that we do not have the ability to measure GFR concomitantly. We currently do not have funding to measure GFR objectively in NOMAS but this is a planned future study if funded. We have acknowledged this is a limitation of our study.

2. Also, the dot plot between eGFRcr (mean, 74.7) and eGFRcys (mean, 51.7) should be presented to know the correlation between the two. As the CKD stage progresses, does the discrepancy become larger?

Response:

We agree and have included the following figure to outline the raw data with a dot and Bland&Altman plot. From our data it seems the discrepancy between eGFRCys and eGFRCr is higher at the higher ranges of GFR estimation than in the lower end. We have included this as the new figure 1 and added a comment in the results section.

Dot plot (left) and the Bland & Altman plot (right) (figure 1)

In results section: “The mean eGFRcr (74.68±18.8 ml/min/1.73m2) was higher than eGFRcys (51.72±17.2 ml/min/1.73m2); there was a greater difference in GFR estimations at the upper rather than lower ranges (figure 1).”

3. It is also important and should be analyzed whether the predicting power of eGFRcys was influenced by the severity of CKD, i.e., CKD stage 1-2 (eGFR > 60), stage 3, or stage 4-5 (eGFR < 30). Mortality incidence should have been more often in advanced CKD stages. This point should be clearly presented.

Response:

We agree with the reviewer than analyzing by CKD stages would have been ideal in our analyses. Unfortunately the proportion of participants with stages 4-5 in our cohort was small and chose to collapse stages 3-5 together. We were nonetheless concerned that severity of CKD would be important and performed our analyses using GFR in a continuous manner (per 10 ml/min/1.73m2) and noted there was a significant association with all-cause mortality.

4. In Figure 3, the ROC curve of eGFRcys seemed much better than eGFRcr among the subjects under 70. Why?

Response:

We thank the reviewer for highlighting this finding in our study. We were concerned that in the older participants in our study serum creatinine would be less predictive due to loss of muscle mass in the elderly, as well as less validated GFR estimation formulae for multi-ethnic populations such as ours.

The discussion has been modified as follows:

“The results, particularly in the participants older than age 70, related to eGFRcys are consistent with findings from other studies suggesting that eGFRcys may be a more accurate estimate of GFR than a serum creatinine-based formula, and extend those findings to an elderly multiethnic population where GFRcr may be confounded by loss of muscle mass which would attenuate the association. The inability to accurately estimate GFR disproportionately affects blacks and Hispanic elderly patients creating significant challenges for prognostication for outcomes, decline of renal function, and management (particularly for medication dosing) of these individuals.”

5. It is not clear what are critical conditions where eGFRcys is superior to eGFRcr in mortality risk prediction. Males? Younger age? Caucasians? Subgroup analysis should be performed in order to specify the factors which have impact in favor of eGFRcys for risk prediction.

Response:

We performed analyses examining GFR estimates by sex, age, and race-ethnicity and noted that for predicting 5 year mortality risk, eGFRcys was better than eGFRcr among those age under 70 years old (p for difference=0.047, compared to age>70) or men (p for difference=0.049, compared to woman). No race-ethnicity differences were found. We have included this in the results section.

For predicting 10 year mortality risk, there were no statistically interactions.

We have included the following table for reference for the reviewer but not in the manuscript since the results were outlined in text.

5 year mortality 10 year mortality

NRI (%) 95% CI of NRI p for difference NRI (%) 95% CI of NRI p for difference

age<70 22.3 (10.6, 34.0) 0.047 4.6 (-2.5,11.8) 0.654

age>=70 9.3 ( 4.2, 14.4) 2.9 ( 0.3, 5.4)

Women 7.4 (1.4, 13.4) 0.049 4.4 ( 1.2 ,7.6) 0.965

Men 16.9 (9.6, 24.3) 4.5 (-0.2, 9.2)

White 8.8 (0.7, 16.9) 0.977 0.8 (-3.5, 5.2) 0.187

Black 12.1 (4.1, 20.0) 6.5 ( 1.9,11.2)

Hispanic 14.9 (6.6, 23.2) 4.9 ( 0.2, 9.7)

We have included the AUC’s in our figures to describe overall model fit in these groups. In conclusion serum cystatin-C based GFR estimated performed better in those under age 70 and in men. Overall however the AUC was low for both cystatin-C and creatinine based GFR estimations emphasizing the need for further better data in diverse populations such as the elderly and women. We have now included this as an additional comment in the results and discussion.

“Interestingly in our cohort the predictive ability of eGFR (regardless of serum measure) appeared higher in the younger participants and men who were most likely to be included in prior cohort that derived GFR estimation formulae. These results highlight the importance of improved accuracy in measurement of GFR in diverse populations will help better understand how CKD is associated with CVD mortality related disparities.”

Reviewer #2:

1. I believe that authors knew serum cystatin C concentration is influenced by many factors, hyper/hypo thyroid, HIV infection, and so on. I could not find in the manuscript about exclusion of these patient from cohort.

Response:

We thank the reviewer for this comment. We did not collect information on thyroid and HIV status is NOMAS, but these conditions were not exclusion criteria for the Northern Manhattan Study. Participants were excluded from a medical condition perspective only if they already had a stroke.

2. Main finding of this study may be “eGFRcys predicted all-cause mortality better than eGFRcre”. However, this fact is already reported indirectly as the authors cited in Ref 24, 33,34, and Astor BC et al 2011, Peralta CA et al 2011

Response:

We agree with the reviewer, however this topic has not been explored to the same degree in diverse, multi-ethnic, and more predominantly older populations such as the Northern Manhattan Study.

Submitted filename: Reviewers comments-PLOS ONE final.docx

Click here for additional data file.

15 Nov 2019

PONE-D-19-20793R1

Creatinine versus Cystatin C for Renal Function-Based Mortality Prediction in an Elderly Cohort: the Northern Manhattan Study

PLOS ONE

Dear Dr Willey,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

As a reviewer pointed out, cystatin C is affected by multiple conditions. It is a limitation of this study that you can not exclude those cohort with thyroid dysfunction, HIV infection and others.  The authors should describe the limitation on this point.

We would appreciate receiving your revised manuscript by Dec 30 2019 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Tatsuo Shimosawa, M.D., Ph.D.

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The revised manuscript by Wiiley et al. responded well to the points raised. I have no further critique.

Reviewer #2: At least authors should refer in manuscript about my previous comment 1. Because it must exist and affect on the results.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

26 Nov 2019

We appreciate the reviewer’s comments on our manuscript. We have included the comments by the reviewers below and have modified the manuscript as requested in the appropriate sections and included below are responses to the reviews.

Reviewer:

1. As a reviewer pointed out, cystatin C is affected by multiple conditions. It is a limitation of this study that you can not exclude those cohort with thyroid dysfunction, HIV infection and others. The authors should describe the limitation on this point.

and

At least authors should refer in manuscript about my previous comment 1. Because it must exist and affect on the results.

Response:

We agree with the reviewer and editor that the lack of this kind of medical comorbidity information is a limitation of our study and have included the following sentence in the limitation sections:

“Cystatin-C levels can be affected by several medical conditions including thyroid dysfunction44 and human immunodeficiency virus infection45 which unfortunately we did not collect in NOMAS.”

We have also added the very helpful references by the reviewer on other studies that have studied creatinine and cystatin as predictors (Astor BC et al 2011, Peralta CA et al 2011).

Submitted filename: Reviewers comments-PLOS ONE v2.docx

Click here for additional data file.

2 Dec 2019

Creatinine versus Cystatin C for Renal Function-Based Mortality Prediction in an Elderly Cohort: the Northern Manhattan Study

PONE-D-19-20793R2

Dear Dr. Willey,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Tatsuo Shimosawa, M.D., Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

2 Jan 2020

PONE-D-19-20793R2

Creatinine versus Cystatin C for Renal Function-Based Mortality Prediction in an Elderly Cohort: the Northern Manhattan Study

Dear Dr. Willey:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Prof. Tatsuo Shimosawa

Academic Editor

PLOS ONE