Confirming the Utility of Four Kidney Biomarker Tests in a Longitudinal Follow-Up Study

Abstract

In this follow-up study, 526 persons were followed for almost 5 years to assess the reversibility and predictive value of four kidney biomarkers in a field epidemiology setting. This study examined (a) whether elevations in urinary albumin, N-acetyl-β-d-glucosaminidase, retinol-binding protein, and alanine aminopeptidase remained elevated at follow-up and (b) whether these initial elevations were predictive of kidney disease (as measured by markers of kidney dysfunction: serum creatinine, serum cystatin C, creatinine clearance, and urine osmolality) at follow-up. Study participants were 8–76 years of age at baseline and were followed for an average of 4.5 years. Approximately 50% of adults who had an elevated biomarker did not have an elevation at followup. Youths with elevated biomarkers at baseline, but who completed adolescence by the time of the follow-up, no longer had any elevations in biomarkers at follow-up. Adult participants who had elevated biomarkers and selected health conditions at baseline (diabetes and, to a lesser extent, heart disease, hypertension, gout, and urinary tract disease) were more likely to show early indicators of kidney impairment at follow-up. Participants with these health conditions and normal kidney biomarker values at baseline had kidney test results at follow-up that were similar to results of study participants who did not have these health conditions at baseline. The presence or absence of elevated biomarkers at baseline among generally healthy participants was not associated with the development of early indicators of kidney impairment at follow-up. This longitudinal study confirmed the utility of these four kidney biomarker tests as markers of preclinical organ dysfunction among adults with certain preexisting medical conditions.

• Alanine aminopeptidase
• Biomarkers
• Albumin
• N-acetyl-β-d-glucosaminidase
• Retinol-binding protein
• Creatinine clearance
• Cystatin C
• Creatinine
• Kidney disease

Introduction

Scientists and community health investigators assessing potential adverse health effects of exposure to hazardous wastes typically begin by collecting data on existing health conditions. Data can be collected through health interviews, medical examinations, laboratory tests, and reviews of medical records. Comparisons can then be made regarding the prevalence or incidence of various illnesses among exposed and unexposed individuals. However, two major obstacles are faced. The latency of some conditions and the potential for certain exposures to make persons more susceptible to future insults mean that studying extant health problems might not answer the question of whether exposed individuals will develop adverse health conditions at a later date. With many data gaps, scientists are often not sure which health end points are related to the exposures at hand. In some cases, the analytic tools available might not be sensitive enough to detect subclinical changes.

To address these concerns, the Agency for Toxic Substances and Disease Registry (ATSDR) used a panel of biomarkers in environmental health investigations to assess clinical and subclinical changes of the kidney. The panel included markers of clinical disease (blood urea nitrogen (BUN) and serum creatinine) as well as research-oriented tests (urinary retinol-binding protein (RBP), urinary alanine aminopeptidase (AAP), urinary N-acetyl-β-d-glucosaminidase (NAG), and quantitative urinary albumin (ALB)).[[1]]

Markers of kidney dysfunction have been reviewed in more detail elsewhere[[2]], [[3]] and are summarized here. Serum creatinine and BUN are standard measures of clinically significant renal changes. Elevated urine albumin (ALB) without an increase in low-molecular-weight (LMW) protein is an early marker of glomerular damage. Increased excretion of the LMW protein RBP is a sensitive indicator of damage to the tubules and loss of tubular protein reabsorption capability.[[1]], [[4]], [[5]] Increased excretion of urinary NAG and AAP signify the direct release of tubular tissue into urine and might indicate tubular cell stimulation, injury, or necrosis. Decreased urine osmolality can be an early indication of damage to the distal tubule and loss of urinary concentrating ability. Microalbuminuria, in particular, has been used as an indicator of kidney impairment in persons with diabetes,[[6]], [[7]], [[8]], [[9]], [[10]] hypertension,[[6]], [[11]], [[12]], [[13]], [[14]], [[15]] and heart disease.[[6]], [[8]], [[12]], [[15]], [[16]], [[17]], [[18]] These markers have also been elevated among workers with nephrotoxic exposures (for example: Refs.[[2]], [[4]], [[5]], [[19]], [[20]], [[21]], [[22]], [[23]], [[24]], [[25]], [[26]]) and in populations exposed to heavy metals.[[27]], [[28]], [[29]] Few studies have followed individuals over time to determine the predictive value of these markers among generally healthy individuals. The interindividual and intraindividual variability in these markers is not well known. Among patients with hypertension, ALB excretion levels can vary greatly over a period of time within individuals with an elevated excretion pattern.[[30]] Among children albumin excretion varies more and is higher than among adults.[[31]]

In 1995, The Joint United States/European Union Workshop “Urinary Biomarkers to Detect Significant Effects of Environmental and Occupational Exposure to Nephrotoxins was held in Atlanta, Georgia.[[5]] This workshop recommended follow-up of individuals with elevated biomarkers in order to obtain information on the predictive ability and reversibility of these biomarkers.[[32]], [[33]] This follow-up study is a result of the workshop's recommendation.

In 1993 and 1994, ATSDR included a panel of urinary kidney biomarkers in three community-based studies in California, Massachusetts, and Nebraska.[[34]], [[35]], [[36]] These studies included communities with groundwater contaminated principally with volatile organic compounds and explosive compounds. Each of these crosssectional health investigations included unexposed comparison communities. At the time of the original studies, kidney biomarkers were not well validated in terms of their predictive value or reversibility, and kidney function in participants in the contaminated communities did not appear to be worse than kidney function in control participants.

The purpose of this follow-up study was to provide individual followup to participants with elevated urinary kidney biomarker tests (and their physicians) and to assess the usefulness of these markers as indicators of preclinical disease. We had two primary research questions. First, are elevations in urinary AAP, NAG, ALB, and RBP stable or reversible over time? Investigators compared participants with an elevated biomarker at baseline with the same participants' follow-up results. Second, are elevations in urinary biomarkers predictive of kidney impairment 5 years later? Investigators compared the likelihood of persons with elevated test results developing an indicator of kidney impairment with the likelihood of persons who did not have an elevated baseline biomarker result developing an indicator of kidney impairment. Indicators of kidney impairment were new elevations in serum creatinine, elevated serum cystatin C, decreased creatinine clearance, and decreased urine osmolality.

Methods

Details on the original, community based studies[[34]], [[35]], [[36]] and follow-up study methods[[37]] have been published elsewhere and are summarized here. More than 2,000 persons participated in the three original studies. Participants answered health questions prefixed with “Has a physician or other health care provider ever told you that you had ______”. A healthy participant was defined as someone who answered “no” at baseline to these questions on “kidney disease”; “diabetes”; “a heart attack, heart disease or other heart problem”; “gout,” or “cancer.” In addition, participants who answered “yes” to ever having been told they had “urinary tract disease” but who answered “no” to “Do you still have it?” were included in the “healthy” category. Female participants were also asked if they were pregnant at the time of the baseline study. The investigators created age-specific and sex-specific 95th percentiles from healthy, nonpregnant participants. The investigators used these percentiles as guidelines to develop cut-points to identify individuals with elevated biomarkers (Table 1). In the case of urinary ALB, investigators were interested in those persons with levels >20 micrograms per gram creatinine (mg/g); therefore, 20 mg/g was used as the entry criteria for urinary ALB. All persons with at least one biomarker above these critical levels were included in the study. Individuals who self-reported cancer, kidney disease, or pregnancy was invited to participate to provide them with individualized follow-up information. These persons were excluded from the analytic portion of the study and their results will not be presented here.

Table 1. Distribution of biomarkers among healthya participants aged 8 through 55 years by sex and selection criteriab for inclusion in the follow-up study

Biomarkerc	Sex	N	Median	5–95th percentile	Selection criteria and definition of “elevated level”b
AAP (U/g)	M	589	4.19	1.49–10.16	11
	F	538	3.74	1.36–7.53	8
NAG (U/g)	M	590	0.64	0.12–1.75	1.8
	F	537	0.72	0.18–1.86	1.9
ALB (mg/g)	M	590	5.0	2.4–21.1	20
	F	539	6.7	3.2–40.9	20
RBP (mg/g)	M	589	0.047	0.011–0.134	0.14
	F	539	0.052	0.011–0.164	0.17

Abbreviations: AAP, alanine aminopeptidase; ALB, albumin; NAG, N-acetyl-β-d-glucosaminidase; RBP, retinol-binding protein.

^aParticipants with urine creatinine levels between 30 and 400 mg/L and who in the original studies self-reported that they (a) never had diabetes, heart disease, hypertension, gout, kidney disease, cancer, or chemotherapy; (b) did not have any urinary tract disease; and (c) were not pregnant.

^bParticipants from the original studies with a baseline biomarker value above the selection criteria were selected for this study and their participation was sought. This same value was used as the definition of “elevated” for statistical analyses.

^cBiomarkers were adjusted for urinary dilution using the ratio to creatinine. All values are given per gram of urine creatinine.

Investigators used frequency matching to obtain appropriate controls matching on age group (8–19, 20–55, and 56–75 years of age at baseline), state (California, Massachusetts, and Nebraska), and general health status. The names of all selected participants older than 50 years of age were compared with the Social Security Death Index to identify individuals who might have died between the baseline and follow-up studies. All individuals not known to be deceased were mailed informational letters and were invited to participate. Data collection took place April–July 1998 at centralized locations in each state. Participants completed a short follow-up health interview; a spot urine sample to repeat ALB, AAP, NAG, RBP, and creatinine; a timed overnight urine sample to calculate creatinine clearance and urine osmolality; and a serum sample to measure creatinine and cystatin C. Baseline and follow-up biomarkers and urine creatinine measurements were performed at the Division of Environmental Health Laboratory Sciences, Centers for Disease Control and Prevention. ALB was measured by enzyme immunosorbent assay as described by Mueller et al.[[38]] An automated Jung measured AAP and Scholz method as described in Mueller et al.[[39]] NAG was measured by the Leaback and Walker method as described in Mueller et al.[[39]] and automated for the Cobas FarrII. RBP was measured using a modification of the method of Topping et al.[[40]] Urine creatinine was measured using a Kodak 250 Analyzer using a single-slide two-point enzymatic method according to manufacturer's directions. Urinary values of ALB, AAP, NAG, and RBP were adjusted for dilution using the ratio to urine creatinine and were expressed for ALB and RBP in mg/g creatinine and for AAP and NAG in U/g creatinine. Two clinical laboratories performed the baseline serum creatinine measurements. A third clinical laboratory performed all follow-up serum creatinine, serum cystatin C, creatinine clearance measurements, and urine osmolality. Cystatin C results were available for participants from 2 of the 3 study sites and values ≥1.5 mg/L were considered “impaired”. The investigators followed references levels as given in Tietz[[41]] to create cutpoints for analysis of serum creatinine and creatinine clearance. Individuals with serum creatinine levels were considered “elevated” as follows: adolescents ages <20 years: >1.0 mg/mL; males ages 18 and older >1.3 mg/mL; females ages 18–60: >1.1 mg/mL; females ages 60 and older: >1.2 mg/dL. Creatinine clearance values <63 mL/min/1.73 m² were considered “impaired”. Timed urine osmolality measurements <300 mosm/kg were considered “impaired”.

Of 2,104 subjects in the original studies, 1,889 had adequate urinary dilution (≥30 mg/dL and <400 mg/dL urinary creatinine). Seventy-eight subjects were excluded from the follow-up study because of cancer or chemotherapy, kidney disease, or pregnancy. At least one elevated biomarker at baseline was found in 397 subjects, and 390 frequency-matched controls were selected. Figure 1 illustrates the subject selection process for the follow-up study.

Figure 1. Selection criteria and participation in the follow-up study.

Results

Six hundred fifty-three persons were approached to participate in the study. Of these, 526 (81%) participated, 112 (17%) refused, and 15 (2%) were no-shows for appointments. Another 134 persons selected were not approached to participate: 28 had died, 50 had moved from the area, 22 could not be contacted, and 34 were lost to follow-up. Figure 1 illustrates the participation rates. Persons who self-reported ever having kidney disease (n = 40) or who had an elevated baseline serum creatinine (n = 27) were excluded from statistical analyses.

One elderly person with an elevated biomarker at baseline developed kidney disease and was on dialysis at the time of follow-up. This person's health was too poor to participate in the study. Most cases of refusal to participate, however, were reported as “too busy” or “not interested” rather than a physical inability to participate.

Stability or Reversibility of Baseline Biomarker Values

To address the first research question, the investigators compared biomarker values in persons with an elevated result at baseline with the follow-up biomarker results in the same persons.

Results of this study varied by age and health status. Among the 56 participants with an elevated biomarker who were 8–19 years of age at baseline, most (76%) were included in the follow-up because of elevated ALB. At follow-up, 62% of these participants with a baseline elevation had reverted to normal, and none of the 32 participants who were 20 years or older at follow-up still had an elevated biomarker (Table 2). Among those who were still under 20 years of age at follow-up, 48% with a baseline elevation were still elevated and 14% of those without a baseline elevation became elevated (Table 2).

Table 2. Comparison of baseline and follow-up biomarker values among participants 8–19 years of age, by age at follow-up.a

Baseline test value	Age <20 years at follow-up new test value		Total	Age ≥20 years at follow-up new test value		Total
	Elevated	Not elevated		Elevated	Not elevated
Any of the four
Elevated	21 (48%)	23 (52%)	44	1b (8%)	12 (92%)	13
Not elevated	5 (14%)	32 (86%)	37	0 (0%)	19 (100%)	19
AAP
Elevated	1 (11%)	8 (89%)	9	0	1	1
Not elevated	3 (8%)	33 (92%)	36	0	18	18
ALB
Elevated	13 (41%)	19 (59%)	32	0	11	11
Not elevated	2 (5%)	35 (95%)	37	0	19	19
NAG
Elevated	1 (11%)	8 (89%)	9	0	1	1
Not elevated	2 (5%)	35 (95%)	37	0	18	18
RBP
Elevated	3 (27%)	8 (73%)	11	0	3	3
Not elevated	2 (5%)	35 (95%)	37	0	19	19

Abbreviations: AAP, alanine aminopeptidase; ALB, albumin; NAG, N-acetyl-β-d-glucosaminidase; RBP, retinol-binding protein.

^aExcludes participants with an elevated serum creatinine or self-reported kidney disease at baseline.

^bThe one new elevation was a newly elevated NAG among a participant with a baseline elevated albumin. At follow-up, the albumin was normal.

Just under half (46%) of persons who were 20–55 years of age at baseline still had an elevated biomarker at follow-up and two-thirds (67%) of those >55 years of age continued to have an elevated biomarker at follow-up. The persistence of elevated biomarkers were highest (91%) among those who had diabetes at baseline, followed by 60% among those with another health condition, and 46% among those without any of these conditions at baseline (Table 3).

Table 3. Comparison of baseline and follow-up biomarker values among participants 20–75 years of age, by baseline health status.a

Baseline test value	Diabetes at baseline				Other disease at baselineb				No disease at baselinec
	New test value				New test value				New test value
	Elevated	Not elevated	Total	p-Valued	Elevated	Not elevated	Total	p-Valued	Elevated	Not elevated	Total	p-Valued
Any of the four
Elevated	10 (91%)	1 (9%)	11	0.180	30 (60%)	20 (40%)	50	0.074	26 (46%)	30 (54%)	56	0.039
Not elevated	4 (57%)	3 (43%)	7	—	33 (38%)	55 (63%)	88	—	16 (26%)	46 (74%)	62	—
AAP
Elevated	2 (40%)	3 (60%)	5	0.317	11 (69%)	5 (31%)	16	0.011	6 (30%)	14 (70%)	20	0.695
Not elevated	1 (14%)	6 (86%)	7	—	17 (20%)	70 (80%)	87	—	12 (20%)	48 (80%)	60	—
ALB
Elevated	5 (71%)	2 (29%)	7	1.000	4 (44%)	5 (56%)	9	0.090	7 (64%)	4 (36%)	11	0.366
Not elevated	2 (29%)	5 (71%)	7	—	12 (14%)	75 (86%)	87	—	7 (11%)	55 (89%)	62	—
NAG
Elevated	6 (60%)	4 (40%)	10	0.414	19 (58%)	14 (42%)	33	0.847	5 (22%)	18 (78%)	23	0.007
Not elevated	2 (29%)	5 (71%)	7	—	13 (15%)	73 (85%)	86	—	5 (8%)	57 (92%)	62	—
RBP
Elevated	2 (40%)	3 (60%)	5	0.317	3 (23%)	10 (77%)	13	0.197	2 (13%)	13 (87%)	15	0.005
Not elevated	1 (14%)	6 (86%)	7	—	5 (6%)	82 (94%)	87	—	2(3%)	60 (97%)	62	—

Abbreviations: AAP, alanine aminopeptidase; ALB, albumin; NAG, N-acetyl-β-d-glucosaminidase; RBP, retinol-binding protein.

^aExcludes participants with an elevated serum creatinine or self-reported kidney disease at baseline.

^bIncludes self-reported hypertension, heart disease, gout, and urinary tract disease.

^cThese persons reported not ever having kidney disease, diabetes, hypertension, heart disease, gout, and urinary tract disease. Other health conditions might have been present.

^dMcNemar's test.

Predictive Value of Elevated Biomarkers for Developing Indicators of Kidney Impairment

We addressed the question of whether elevations in urinary biomarkers are predictive of kidney impairment 5 years later by comparing the likelihood of persons with elevated test results developing an indicator of kidney impairment with those persons who did not have an elevated baseline biomarker result. Indicators of kidney impairment are new elevation in serum creatinine, elevated serum cystatin C, decreased creatinine clearance, and decreased urine osmolality. No significant differences in mean kidney measures at followup existed among participants from “exposed” and “unexposed” comparison communities and their results were pooled. Study results varied by age and health status at baseline. Among participants <20 years of age, baseline elevations in biomarkers were not related to any of the markers of kidney impairment. Results for participants 20–76 are provided in the following paragraphs.

Association Between Baseline Biomarker Values and Follow-Up Serum Creatinine

Elevated Serum Creatinine

Overall, 2% (i.e., 6 of 322 participants 20–76 years of age at baseline without self-reported kidney disease or elevated serum creatinine at baseline) of participants developed newly elevated serum creatinine. The risk for elevated serum creatinine was higher (4%) among those who had any of the four biomarkers elevated at baseline compared to those with no elevated biomarkers (1%) at baseline (p = 0.051). The risk was highest for those persons with an elevated baseline biomarker value who also had diabetes at baseline (15%). This risk was primarily in the older participants (Table 4) and not significant in those ages 20–55 or those who were healthy at baseline. Among adults 56–75 years of age, having an elevated AAP, ALB, or NAG was significantly associated with developing an elevated serum creatinine at follow-up (Tables 4 and 5). These elevated serum creatinines occurred primarily among those with diabetes at baseline (Table 5). The data were too sparse to adjust for potential confounding effects of age, sex, or state.

Table 4. Risk of having elevated serum creatinine, low creatinine clearance, and elevated serum cystatin C values at follow-up, by baseline biomarker status among participantsa 56–75 years of age

	High serum creatinine		Low creatinine clearance		High serum cystatin C
Baseline biomarker value	No. with high serum creatinine (total)	Risk ratiob	No. with creatinine clearance <63 mL (total)	Risk ratio	No. with elevated cystatin C (>1.5 mg.dL) (total)	Risk ratio
Any of the four biomarkers elevated	4 of 69 (6%)	NC, p = 0.030	13 of 63 (21%)	2.5 (1.2–5.2), p = 0.049	6 of 42 (14%)	8.6 (1.5–49.1), p = 0.019
Elevated ALB	2 of 16 (13%)	NC, p = 0.020	5 of 14 (36%)	4.3 (1.9–9.9), p = 0.013	4 of 10 (40%)	24.0 (4.2–138.3), p = 0.001
Elevated AAP	3 of 29 (10%)	NC, p = 0.012	4 of 26 (15%)	1.8 (0.7–4.8), p = 0.285	2 of 15 (13%)	8.0 (1.1–56.7), p = 0.100
Elevated NAG	5 of 45 (11%)	NC, p = 0.003	14 of 43 (33%)	3.6 (1.8–7.4), p = 0.003	6 of 32 (19%)	11.3 (2.0–64.1), p = 0.007
Elevated RBP	1 of 14 (7%)	NC, p = 0.128	2 of 13 (15%)	1.8 (.5–6.3), p = 0.346	3 of 10 (30%)	18.0 (2.9–110.5), p = 0.008
None elevated	0 of 95 (0%)	Reference	8 of 85 (9%)	Reference	1 of 60 (2%)	Reference

Abbreviations: AAP, alanine aminopeptidase; ALB, albumin; NAG, N-acetyl-β-d-glucosaminidase; NC, cannot be computed; RBP, retinol-binding protein.

^aExcludes participants with self-reported kidney disease at baseline or elevated baseline serum creatinine. Sample sizes vary due to missing data.

^b90% confidence interval (logit); Fisher's exact 2-tailed p-value.

Table 5. Risk of having elevated serum creatinine values at follow-up, by baseline biomarker status and baseline health status, among participantsa 20–75 years of age

	Diabetes		Other diseaseb		No diseasec
Baseline biomarker value	No. with high serum creatinine (total)	Risk ratiod	No. with high serum creatinine (total)	Risk ratio	No. with elevated high serum creatinine (total)	Risk ratio
Any of the four biomarkers elevated	2 of 13 (15%)	NC, p = 0.505	2 of 62 (3%)	NC, p = 0.149	1 of 67 (1%)	1.1 (0.1–11.1), p = 1.000
Elevated ALB	2 of 7 (29%)	NC, p = 0.200	0 of 12 (0%)	No difference	0 of 13 (0%)	0, p = 1.00
Elevated AAP	1 of 6 (17%)	NC, p = 0.429	1 of 22 (5%)	NC, p = 0.192	1 of 24 (4%)	3.0 (0.3–29.4), p = 0.443
Elevated NAG	2 of 11 (18%)	NC, p = 0.485	2 of 42 (2%)	NC, p = 0.090	1 of 31 (3%)	2.4 (0.2–23.8), p = 0.505
Elevated RBP	1 of 4 (25%)	NC, p = 0.333	1 of 12 (8%)	NC, p = 0.109	0 of 15 (0%)	0, p = 1.00
None elevated	0 of 8 (0%)	Reference	0 of 98 (0%)	Reference	1 of 74 (1%)	Reference

Abbreviations: AAP, alanine aminopeptidase; ALB, albumin; NAG, N-acetyl-β-d-glucosaminidase; NC, cannot be computed; RBP, retinol-binding protein.

^aExcludes participants with self-reported kidney disease at baseline or elevated baseline serum creatinine.

^bIncludes self-reported hypertension, heart disease, gout, and urinary tract disease.

^cThese persons reported not ever having kidney disease, diabetes, hypertension, heart disease, gout, and urinary tract disease. Other health conditions might have been present.

^d90% confidence interval (logit); Fisher's exact 2-tailed p-value.

Mean Serum Creatinine

Mean serum creatinine levels at follow-up, adjusted for age, sex, and state of data collection, were examined in participants 56–75 years of age. Participants in this age group appeared most at risk for kidney impairment. Among diabetics ages 56–75 years of age, adjusted mean serum creatinine levels were higher for participants with baseline elevations of AAP, ALP, NAG, and RBP compared to without an elevated biomarker. No differences in mean serum creatinine levels were found between persons with and without elevated baseline biomarker values among those with other health conditions or no health conditions at baseline.

Association Between Baseline Biomarker Values and Creatinine Clearance

Low Creatinine Clearance Values at Follow-Up

Overall, 29 (10%) of 280 participants with information on creatinine clearance who reported never having kidney disease at baseline and who did not have an elevated serum creatinine at baseline had a low creatinine clearance at follow-up. Thirty-six percent of participants over age 55, 44% of diabetics, and 25% of participants with other health problems had low creatinine clearance values at follow-up. Participants >20 years of age were two to three times more likely to have a decreased creatinine clearance if they started with an elevated biomarker compared to those having a lower biomarker value at baseline. ALB and NAG were the most predictive of the four biomarkers for ages 56–75 years (Table 4). Participants with an elevated urine ALB or NAG at baseline and who also were either >56 years of age (Table 4) or had diabetes or another health problem at baseline (Table 6) were two to three times as likely to have a low creatinine clearance at follow-up as those without elevated biomarkers at baseline. For adults not reporting diabetes or other health conditions an elevated NAG was associated with the greatest risk for low creatinine clearance at follow-up (Table 6). Participants with baseline urine ALB >200 were the most likely to have a low creatinine clearance at baseline, particularly if they were >55 years of age, diabetic, or had another health problem.

Table 6. Risk of having a low creatinine clearance value (<63 mL/min) at follow-up among participantsa 20–75 years of age, by baseline health status

	Diabetes		Other diseaseb		No diseasec
Baseline biomarker value	No. with low creatinine clearance (total)	Risk ratiod	No. with low creatinine clearance (total)	Risk ratio	No. with low creatinine clearance (total)	Risk ratio
Any of the biomarkers	4 of 14 (29%)	NC, p = 0.254	11 of 54 (20%)	1.9 (1.0–3.8), p = 0.138	4 of 57 (7%)	4.4 (0.7–26.7), p = 0.192
ALB	4 of 9 (44%)	NC, p = 0.082	2 of 8 (25%)	2.4 (0.8–7.3), p = 0.209	0 of 8 (0%)	0, p = 1.00
AAP	3 of 7 (43%)	NC, p = 0.077	2 of 18 (11%)	1.0 (0.3–3.5), p = 1.00	1 of 21 (5%)	3.0 (0.3–29.1), p = 0.444
NAG	4 of 12 (33%)	NC, p = 0.117	9 of 37 (24%)	2.6 (1.2–5.3), p = 0.033	4 of 29 (14%)	8.6 (1.4–51.8), p = 0.034
RBP	3 of 6 (50%)	NC, p = 0.055	1 of 10 (10%)	1.05 (0.2–5.5), p = 1.00	0 of 14 (0%)	0, p = 1.00
None	0 of 8 (0%)	Reference	9 of 85 (11%)	Reference	1 of 62 (2%)	Reference

Abbreviations: AAP, alanine aminopeptidase; ALB, albumin; NAG, N-acetyl-β-d-glucosaminidase; NC, cannot be computed; RBP, retinol-binding protein.

Summary, elevated NAG and elevated albumin were the most predictive of having a lower creatinine clearance at follow up than the other two biomarkers.

^aExcludes participants with self-reported kidney disease at baseline, elevated baseline serum creatinine, or problems with urine collection.

^bIncludes self-reported hypertension, heart disease, gout, and urinary tract disease.

^cThese persons reported not ever having kidney disease, diabetes, hypertension, heart disease, gout, and urinary tract disease. Other health conditions might have been present.

^d90% confidence interval from logit estimator; Fisher's exact 2-tailed p-value.

Mean Creatinine Clearance Values at Follow-Up

Mean creatinine clearance values, adjusted for age, sex, and state of data collection, were compared by baseline health status. For participants <56 years of age, no differences were observed in adjusted mean creatinine clearance values for those with an elevated baseline biomarker vs. those with a normal baseline biomarker value for AAP, NAG, ALB, and RBP, regardless of baseline health status. For participants who were >55 years of age and had no health problems, creatinine clearance values at follow-up were similar between those with elevated or normal biomarker values. However, participants >55 years of age who also had a co-morbid health condition had lower mean creatinine clearance values if any of the biomarkers were elevated. Baseline ALB level had the greatest impact with

• Those persons with a co-morbid condition and a baseline urinary ALB level >200 milligrams per gram (mg/g) having a mean creatinine clearance value of 50.4 milliliters per minute (mL/min),

• Those persons with levels 20–199 mg/g having a value of 74.7 mL/min, and

• Those persons with urine ALB levels <20 having a value of 101.5 mL/min.

Diabetics who had an elevated biomarker value had the lowest creatinine clearance values at follow-up. However, diabetics with normal biomarker values had similar creatinine clearance values, as did participants without any health conditions or without elevated biomarker values.

Association Between Baseline Biomarkers and Serum Cystatin C at Follow-Up

An excess risk of having an elevated serum cystatin C level due to having an elevated urinary biomarker was observed for those persons >55 years of age and those with an underlying health problem (Tables 4 and 7).

Table 7. Risk of having an elevated serum cystatin C value (>1.5 mg/dL) at follow-up among participants 20–75 years of age, by baseline health status

	Diabetes		Other disease		No disease
Baseline biomarker value	No. with low elevated cystatin (total)	Risk ratioc	No. with low elevated cystatin (total)	Risk ratioc	No. with low elevated cystatin (total)	Risk ratioc
Any	2 of 10 (20%)	NC, p = 0.500	5 of 38 (13%)	8.2 (14–47.9), p = 0.028	2 of 49 (4%)	NC, p = 0.228
ALB	2 of 6 (33%)	NC, p = 0.455	2 of 9 (22%)	13.8 (2.0–94.7), p = 0.041	0 of 10 (0%)	NC, no difference
AAP	1 of 5 (20%)	NC, p = 0.455	2 of 13 (15%)	9.5 (1.4–67.1), p = 0.076	1 of 20 (10%)	NC, p = 0.274
NAG	2 of 9 (22%)	NC, p = 0.486	5 of 28 (18%)	11.1 (1.9–64.5), p = 0.010	2 of 24 (8%)	NC, p = 0.094
RBP	2 of 5 (40%)	NC, p = 0.182	1 of 8 (13%)	7.8 (0.8–73.0), p = 0.217	1 of 9 (11%)	NC, p = 0.145
None	0 of 6 (0%)	Reference	1 of 62 (2%)	Reference	0 of 53 (0%)	Reference

Abbreviations: AAP, alanine aminopeptidase; ALB, albumin; NAG, N-acetyl-β-d-glucosaminidase; NC, cannot be computed; RBP, retinol-binding protein.

^a90% Confidence intervals from logit estimator; Fisher's exact 2-tailed p-value.

^aIncludes self-reported hypertension, heart disease, gout, and urinary tract disease.

^bThese persons reported not ever having kidney disease, diabetes, hypertension, heart disease, gout, and urinary tract disease. Other health conditions might have been present.

^c90% Confidence intervals (logit); Fisher's exact 2-tailed p-value.

Among those persons assessed for the effects of urine ALB, no one <55 years of age or without a baseline health condition had an elevated serum cystatin C at follow-up. Among those 56–75 years of age, significant risk ratios between 11.3 and 24.0 (for albumin) were observed for those with an elevated albumin, NAG, and RBP for an elevated cystatin C at follow-up (Table 4). Among those with elevated urinary ALB levels, an excess risk of having elevated serum cystatin C occurred among those >56 years of age and those who had diabetes or hypertension, heart disease, gout, or urinary tract disease. Similarly, excess risk of having an elevated serum cystatin C at follow-up for those with a baseline elevated NAG was limited to participants who were >55 years of age and those with diabetes or another chronic health condition. Among participants 56–75 years of age with a baseline health condition, adjusted mean cystatin C values were higher for those with any of the elevated biomarkers compared with those with a normal biomarker value. No differences were observed for those without a baseline health problem.

Association Between Baseline Biomarkers and Urine Osmolality at Follow-Up

Having an elevated baseline urinary biomarker was not associated with having a low timed urine osmolality at follow-up. No one with an ALB >200 mg/g had a low osmolality at follow-up. Urinary osmolality was not associated with diabetes. None of the 17 participants with diabetes and an elevated biomarker had a low osmolality at follow-up.

Discussion

This study, unlike clinical studies, was able to examine the effects of preclinical biomarkers of early kidney damage in a large number of generally healthy, nonhospitalized people in a wide range of ages. This study benefits from observing kidney function changes over time in a cross-section of a number of American communities. Further, 80% of persons successfully contacted agreed to participate-an acceptably high participation rate for a community-based study.

Little is known about the interpretation of markers of kidney impairment in healthy people <20 years of age. The interpretation of biomarkers in children and adolescents is different than that in adults.[[42]] Participants who were adolescents at baseline but 20 years of age or older at follow-up had almost no elevated biomarkers compared to the 40% found in the group <20 years of age at follow-up (the majority were still going through puberty). These findings agree with findings from other populations that found puberty-related biomarker changes.[[31]] Clinical indicators of renal damage (e.g., elevated serum creatinine, cystatin C, decreased creatinine clearance, or urine osmolality) were rare in this population. This gives further evidence that these biomarkers could be more related to changes associated with puberty than with kidney damage in those under age 20 years.

A certain degree of intraindividual variation occurs in biologic measurements. The intraindividual variability for urine ALB is quite large.[[30]] Most people do not have elevated albumin excretion patterns and have low intraindividual variance. However, those with an elevated excretion pattern have approximately 100-fold greater intraindividual variance.[[30]] RBP has a mean intraindividual variability of 40% in spot samples taken over a week and 50% over an 8-year follow-up period.[[43]] In what has been described as regression to the mean (termed “regression fallacy” by DeGroot[[44]]), it is possible that some persons who initially had a biomarker elevation would no longer have this elevation. Just as variability occurred in estimating the “true” baseline biomarker value, variability also occurred in the single follow-up measurement. This study has provided information on the long-term (up to 5-year) intraindividual variability of people either with or without certain health conditions. Future studies measuring short-term and longer-term intraindividual variability would be useful, particularly studies that measure this variability among generally healthy individuals.

The use of hospital-based tests, such as magnetic resonance imaging or 24-h glomerular filtration rate based on supervised urine collection for a full 24 h, would have provided more accurate and reliable kidney function information. However, these tests are time-consuming and invasive, which would have made them unacceptable for the field epidemiology setting. Therefore, the use of an overnight urine collection to estimate creatinine clearance was a useful addition to the set of kidney function tests. Although it was anticipated that this collection would be difficult to do logistically, 85% of participants successfully provided the needed specimens.

A different laboratory was used for performing the standard clinical tests at baseline and at follow-up. Differences in laboratories could mean that comparisons of baseline and follow-up serum creatinine could be less accurate than if they had they been from the same laboratory. However, adjustment was made for study location, which somewhat adjusted for the different laboratories used. The same laboratory (National Center for Environmental Health, Centers for Disease Control and Prevention) was used for baseline and follow-up kidney biomarker tests.

Conclusions

Among generally healthy study participants, having an elevated biomarker carried no excess kidney health risk after 4.4 years of followup. Among participants with diabetes and, to a lesser extent, heart disease, hypertension, gout, and urinary tract disease, having an elevated biomarker was a risk factor for developing an elevated serum creatinine, elevated serum cystatin C, and decreased creatinine clearance after 4.4 years of follow-up compared to those who did not have an elevated biomarker at baseline.

Participants who reported having these health conditions with normal biomarker levels at baseline had similar kidney test results at follow-up as did people without the health conditions. We did not find urine osmolality to be associated with any of the biomarkers at follow-up. Finally, biomarker elevations among adolescents disappeared among those who had completed puberty. Elevations in participants 8–19 years of age were not associated with markers of kidney impairment at follow-up and appear to be related to changes in adolescence because none of those who had matured out of the adolescent stage had an elevated biomarker. Elevated biomarker values were most stable among those who were older or who had extant health conditions.

Abbreviations
AAP:	=	Alanine aminopeptidase
ALB:	=	Albumin
ATSDR:	=	Agency for Toxic Substances and Disease Registry
BUN:	=	Blood urea nitrogen
LMW:	=	Low molecular weight
NAG:	=	N-Acetyl-β-d-glucosaminidase
NC:	=	Cannot be computed
RBP:	=	Retinol-binding protein