Circulating Plasma Biomarkers in Biopsy-Confirmed Kidney Disease

Visual Abstract

Abstract

Background and objectives

Biomarkers for noninvasive assessment of histopathology and prognosis are needed in patients with kidney disease.

Design, setting, participants, & measurements

Using a proteomics assay, we measured a multimarker panel of 225 circulating plasma proteins in a prospective cohort study of 549 individuals with biopsy-confirmed kidney diseases and semiquantitative assessment of histopathology. We tested the associations of each biomarker with histopathologic lesions and the risks of kidney disease progression (defined as ≥40% decline in eGFR or initiation of KRT) and death.

Results

After multivariable adjustment and correction for multiple testing, 46 different proteins were associated with histopathologic lesions. The top-performing markers positively associated with acute tubular injury and interstitial fibrosis/tubular atrophy were kidney injury molecule-1 (KIM-1) and V-set and Ig domain-containing protein 2 (VSIG2), respectively. Thirty proteins were significantly associated with kidney disease progression, and 35 were significantly associated with death. The top-performing markers for kidney disease progression were placental growth factor (hazard ratio per doubling, 5.4; 95% confidence interval, 3.4 to 8.7) and BMP and activin membrane-bound inhibitor (hazard ratio, 3.0; 95% confidence interval, 2.1 to 4.2); the top-performing markers for death were TNF-related apoptosis-inducing ligand receptor-2 (hazard ratio, 2.9; 95% confidence interval, 2.0 to 4.0) and CUB domain-containing protein-1 (hazard ratio, 2.4; 95% confidence interval, 1.8 to 3.3).

Conclusion

We identified several plasma protein biomarkers associated with kidney disease histopathology and adverse clinical outcomes in individuals with a diverse set of kidney diseases.

Podcast

This article contains a podcast at https://www.asn-online.org/media/podcast/CJASN/2021_12_28_CJN09380721.mp3

Introduction

The two measures used to diagnose and stage kidney diseases, albuminuria and eGFR, do not provide specificity about kidney histopathologic lesions and are imperfect prognostic markers. Novel biomarkers with greater specificity for histopathologic lesions will enhance clinical phenotyping of kidney diseases and may provide an important milestone toward precision medicine.

Several biomarkers of kidney inflammation and tubular injury have been shown to prognosticate the risk of adverse clinical outcomes in patients with kidney diseases (1–4), but these biomarkers only represent a small segment of the human blood proteome. Proteomics-based approaches add to this by allowing for an unsupervised protein discovery that does not entirely depend on preexisting pathophysiologic knowledge. The history of biomarker studies in other clinical disciplines illustrates how large-scale proteomics approaches can identify important protein markers for improved risk prediction (5) and the development of new therapeutic agents (6). Fewer studies, however, have made use of these techniques for investigation of human kidney disease (7–9).

In this study, we used a novel proteomics assay to measure a multimarker panel of 225 plasma proteins in a prospective cohort study of individuals with biopsy-confirmed kidney diseases and adjudicated semiquantitative assessment of histopathology. We tested the associations of each biomarker with kidney histopathologic lesions and the risks of subsequent kidney disease progression and death and explored pathways with potential biologic relevance for patients at risk of kidney disease progression and death using the Reactome pathway database.

Materials and Methods

Study Population

The Boston Kidney Biopsy Cohort (BKBC) is a prospective, observational cohort study of adult patients undergoing native kidney biopsy between September 2006 and October 2018 at three tertiary care hospitals in Boston, Massachusetts, including Brigham and Women’s Hospital, Massachusetts General Hospital, and Beth Israel Deaconess Medical Center. Details of the study design have been previously described (10). Patients provided blood and urine samples on the day of kidney biopsy. For this study, we evaluated 549 participants with available plasma samples that included both common forms (kidney disease attributed to diabetes or hypertension) and more rare forms of kidney diseases (all other etiologies, such as other glomerular, tubulointerstitial, or vascular diseases). The Partners Human Research Committee (the Brigham and Women’s Hospital Institutional Review Board) approved the study protocol, which is in accordance with the principles of the Declaration of Helsinki.

Sample Collection, Proteomics Assays, and Exposures

Blood samples were collected from study participants on the day of biopsy, aliquoted, and immediately stored at −80°C. Aliquots were analyzed at Olink using high-throughout, multiplex immunoassays (11) on three commercially available panels named Inflammation, Organ Damage, and Cardiovascular II. Each panel consists of 92 biomarker proteins that were chosen on the basis of their potential relevance in various pathologic processes. All protein values are expressed as normalized protein expression values on a log₂ scale. We included 5% blind split replicates in addition to BKBC samples. Of the 276 proteins included in the three panels, we used 225 biomarkers as the primary exposures for statistical analyses that were nonoverlapping across the panels and passed our quality control metrics (Supplemental Material).

Histopathologic Outcomes

Methods to evaluate and score histopathologic lesions were previously described in detail (10). Kidney biopsy specimens were adjudicated under light microscopy by two experienced kidney pathologists who provided semiquantitative scores of kidney inflammation, fibrosis, vascular sclerosis, and acute tubular injury (ATI). Global and segmental glomerulosclerosis, endocapillary and extracapillary glomerular inflammation, focal glomerular necrosis, fibrocellular crescents, interstitial fibrosis/tubular atrophy (IFTA), and inflammation in the fibrosed and nonfibrosed interstitium were scored from zero to three, reflecting the percentage of affected total cortical volume of ≤10%, 11%–25%, 26%–50%, and >50%, respectively. Mesangial expansion, ATI, and arterial and arteriolar sclerosis were scored from zero to three, reflecting none, mild, moderate, and severe lesions, respectively (Supplemental Table 1). Additional details on evaluation and scoring of histopathologic lesions are provided in Supplemental Material. All participants’ charts were reviewed alongside histopathologic evaluations to provide the final primary clinicopathologic diagnosis.

Clinical Outcomes

The primary outcome was kidney disease progression, defined as ≥40% decline in eGFR or kidney failure (defined as the initiation of dialysis or kidney transplantation). The secondary outcome was death. To ascertain information on vital status, change in creatinine, or need for dialysis, we reviewed the electronic medical record (EMR) of the respective hospital as well as other linked EMR systems. This review consisted of reviewing progress notes, discharge summary, laboratory tests, and other documentation in EMR. Data on eGFR during follow-up were obtained from the EMR, and KRT status was confirmed by reviewing the EMR and linkage with the United States Renal Data System database. Mortality status was confirmed with the Social Security Death Index. Participants were followed until the occurrence of death, voluntary study withdrawal, loss to follow-up (determined by the absence of clinical information during follow-up), or February 1, 2020.

Covariates

Detailed patient information was collected at the biopsy visit, including demographics, medical history, medication lists, and pertinent laboratory data. We obtained serum creatinine (SCr) from the EMR on the day of biopsy. In participants for whom this was unavailable, we measured SCr in available blood samples collected on the day of biopsy. We obtained spot urine protein-creatinine ratio (UPCR) or urine albumin-creatinine ratio (UACR) from the date of kidney biopsy to 3 months before biopsy from the EMR. If both were available, UACR was used. If a participant did not have any of these values, we measured UACR from urine collected on the day of the kidney biopsy. SCr and urine creatinine were measured using a Jaffe-based method, and urine albumin was measured by an immunoturbidometric method. The creatinine-based Chronic Kidney Disease Epidemiology Collaboration equation was used to calculate the eGFR (12).

Pathway Analyses

To obtain basic functional information on our proteins of interest and to investigate potentially relevant biologic pathways, we used Reactome, an open-source database for pathway analyses (13) for biomarkers that were significantly associated with kidney disease progression and death after correction for multiple testing. Reactome uses the Benjamini–Hochberg approach to provide a false discovery rate that accounts for the number of tests performed and ranks pathways on the basis of P values calculated using a permutation approach that permutes group labels using 1000 permutations as well as the false discovery rate.

Statistical Analyses

We summarized descriptive statistics as count with percentage for categorical variables and mean±SD or median with interquartile range for continuous variables. For skewed data distributions, we performed natural logarithmic transformation as appropriate. Unadjusted and adjusted multivariable linear regression models were used to assess associations of each plasma biomarker protein with histopathologic lesions. Histopathologic lesions were dichotomized as described in Supplemental Table 1. Percentage differences of plasma biomarkers by histopathologic lesion were calculated by raising two to the power of the β-coefficient, subtracting one, and multiplying by 100 [(2^β−1)×100)] for each respective histopathologic lesion. We limited statistical analyses on histopathologic lesions to participants with adjudicated histopathology by both kidney pathologists (n=411; 75%), except for analyses of global or segmental glomerulosclerosis because they were taken from the biopsy report (n=549).

We performed time-to-event analyses to examine the associations of biomarkers with kidney disease progression and death. Cox proportional hazards models using kidney disease progression or death as the dependent variable were stratified by site and adjusted for age, race, sex, log(proteinuria), eGFR (modeled continuously), and primary clinicopathologic diagnostic category of kidney disease. For the outcome of kidney disease progression, we treated the data as interval censored because the exact date of the event may not be known. We evaluated the association between plasma biomarkers and subsequent kidney disease progression using a nonparametric survival function for interval-censored data (14). We confirmed no violations of the proportional hazards assumption through assessment of Schoenfeld residuals and used complete case analysis for the analyses as there were <5% missing data. To counteract the problem of multiple testing, we used Bonferroni correction for all reported P values (adjusted P value = 225 × P_uncorrected). Statistical analyses were performed using R Version 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Baseline Characteristics

Baseline characteristics of the study cohort are shown in Table 1. The mean age was 52±17 years, and 52% were women. The mean eGFR was 58±36 ml/min per 1.73 m², and the median proteinuria (interquartile range) was 1.7 (0.4–3.8) g/g creatinine. The most common primary clinicopathologic diagnoses were glomerulopathies (47%), diabetic nephropathy (12%), advanced glomerulosclerosis (11%), vascular disease (9%), and tubulointerstitial disease (9%).

Table 1.

Baseline characteristics of participants in the Boston Kidney Biopsy Cohort

Baseline Characteristics	Boston Kidney Biopsy Cohort Participants, n=549
Clinical characteristics
Age, yr	52±17
Women	288 (52)
Race
Black	105 (19)
White	347 (63)
Other	97 (18)
eGFR, ml/min per 1.73 m2	58±36
eGFR categories
Category 1 (eGFR>90 ml/min per 1.73 m2)	132 (24)
Category 2 (eGFR=60–89 ml/min per 1.73 m2)	92 (17)
Category 3A (eGFR=45–59 ml/min per 1.73 m2)	81 (15)
Category 3B (eGFR=30–44 ml/min per 1.73 m2)	91 (17)
Category 4 (eGFR=15–29 ml/min per 1.73 m2)	101 (18)
Category 5 (eGFR<15 ml/min per 1.73 m2)	52 (9)
Proteinuria, g/g creatinine	1.7 [0.4–3.8]
Proteinuria categories, g/g creatinine
Category 1, <0.15	73 (13)
Category 2, 0.16–1	139 (25)
Category 3, 1.1–3.5	175 (32)
Category 4, >3.5	162 (30)
Reason for biopsy
Proteinuria	312 (57)
Hematuria	134 (24)
Nephrotic syndrome	69 (13)
Nephritic syndrome	14 (3)
Abnormal eGFR/other	336 (61)
Primary clinicopathologic diagnosis
Proliferative glomerulonephritis	162 (30)
Nonproliferative glomerulopathy	96 (17)
Diabetic nephropathy	64 (12)
Advanced glomerulosclerosis	62 (11)
Vascular disease	52 (9)
Tubulointerstitial disease	47 (9)
Paraprotein-related disease	35 (6)
Other	31 (6)
Comorbid conditions
Diabetes mellitus	122 (22)
Hypertension	287 (52)
SLE	90 (16)
Hepatitis C	11 (2)
Hepatitis B	4 (1)
HIV	6 (1)
Malignancy	80 (15)
Medications
ACEi/ARB	257 (47)
MRA	14 (3)
Calcium channel blockers	137 (25)
β-blockers	166 (30)
Immunosuppression	102 (19)
Corticosteroids	123 (22)

Data are presented as mean±SD, median [interquartile range], or count with frequency (percentage) for binary and categorical variables. Data on proteinuria were missing for seven individuals. Percentages do not add to 100 as there may have been more than one reason for a kidney biopsy. ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; MRA, mineralocorticoid receptor blocker.

Plasma Protein Biomarkers Associated with Histopathologic Lesions

Associations between plasma biomarkers and histopathologic lesions are shown in Figure 1 and Table 2. After multivariable adjustment and correction for multiple testing, 46 different plasma proteins (57 associations total) were independently associated with different histopathologic lesions. By level of statistical significance, the top-performing biomarkers positively associated with more severe ATI and IFTA were kidney injury molecule-1 (KIM-1) and V-set and Ig domain-containing protein 2 (VSIG2), respectively (Table 2). The top-performing biomarkers positively associated with other lesions included protein S100-A12 (EN-RAGE) with greater severity of glomerular inflammation, C-X-C motif chemokine 9 with inflammation in the nonfibrosed interstitium, tissue factor with glomerular sclerosis, and TNF-related apoptosis-inducing ligand (TRAIL) with mesangial expansion. Matrix metalloproteinase-7 and fatty acid–binding protein 2 (FABP2) were the top-performing markers associated with more severe inflammation in the fibrosed interstitium and arteriolar sclerosis, respectively. Seven different biomarkers (nine associations total) were inversely associated with histopathologic lesions (Table 2). Among those, stem cell factor (SCF) associated inversely with both ATI and inflammation in the nonfibrosed interstitium. There was no significant association between any of the biomarkers and segmental glomerulosclerosis or arterial sclerosis after multivariable adjustment.

Figure 1.

Circulating plasma proteins associated with histopathologic lesions. The heat map shows associations between histopathologic lesions and biomarkers after Bonferroni correction. β-coefficients are derived from multivariable linear regression models adjusted for age, sex, race, and eGFR (modeled continuously), and they are displayed as colors ranging from blue to red. Reference is none/mild lesion severity for acute tubular injury and arteriolar sclerosis. Reference is absence of lesion for mesangial expansion. Reference is 0%–10% of cortical volume or glomeruli affected for inflammation in the fibrosed and nonfibrosed interstitium and glomerular inflammation. Reference is 0%–25% of cortical volume or glomeruli affected for glomerular sclerosis and interstitial fibrosis/tubular atrophy (IFTA). Endocapillary glomerular inflammation, extracapillary cellular crescents, focal glomerular necrosis, and fibrocellular crescents were combined into a single dichotomous variable named “glomerular inflammation.”

Table 2.

Associations of histopathologic lesions with plasma biomarkers

Plasma Protein	Percentage Difference in Plasma Protein Associated with Histopathologic Lesion (95% Confidence Interval)a	P Value
Glomerular inflammation
EN-RAGE	98 (61 to 144)	<0.001
DNER	−15 (–20 to –10)	<0.001
MCP-3	51 (30 to 77)	<0.001
NOS3	46 (24 to 71)	0.002
Gal-9	19 (10 to 28)	0.003
IL-1ra	35 (18 to 55)	0.005
CSF-1	11 (5 to 16)	0.02
IL-6	48 (21 to 80)	0.03
GDF-2	−23 (–32 to –12)	0.03
CCL4	32 (15 to 52)	0.03
hOSCAR	9 (4 to 13)	0.03
DPP6	−14 (–20 to –7)	0.04
CXCL10	53 (23 to 90)	0.04
PDCD1	16 (7 to 25)	0.05
Inflammation, nonfibrosed interstitium
SCF	−22 (–31 to –13)	0.003
CXCL9	72 (34 to 120)	0.01
MMP-7	22 (11 to 33)	0.01
DPP6	−17 (–24 to –9)	0.02
CCL23	24 (11 to 38)	0.04
CALCA	55 (23 to 95)	0.04
Glomerular sclerosis
TF	17 (11 to 24)	<0.001
PRSS27	24 (14 to 34)	<0.001
SCF	23 (14 to 32)	<0.001
GIF	43 (25 to 65)	<0.001
CA14	18 (9 to 28)	0.01
PTX3	−17 (–24 to –9)	0.01
GDF-2	23 (11 to 35)	0.02
VSIG2	21 (10 to 32)	0.02
IL-10	−20 (–28 to –10)	0.04
NPPC	25 (11 to 41)	0.04
Mesangial expansion
TRAIL	20 (10 to 30)	0.01
Gal-9	22 (11 to 34)	0.01
CSF-1	13 (7 to 20)	0.01
ENTPD6	14 (7 to 21)	0.02
PD-L2	19 (9 to 29)	0.02
TRANCE	42 (19 to 70)	0.03
ADM	24 (11 to 39)	0.03
CCL11	24 (11 to 39)	0.04
Interstitial fibrosis/tubular atrophy
VSIG2	30 (16 to 46)	0.003
TF	17 (9 to 26)	0.01
AMBP	10 (5 to 15)	0.01
IL-10RB	13 (7 to 20)	0.01
PGF	19 (9 to 29)	0.02
FGF-23	47 (22 to 79)	0.02
CLEC1A	23 (11 to 36)	0.02
PTK7	22 (10 to 35)	0.03
PAR-1	17 (8 to 27)	0.03
TNFRSF-9	33 (15 to 54)	0.04
Inflammation, fibrosed interstitium
MMP-7	21 (10 to 33)	0.02
SPON2	8 (4 to 13)	0.03
KIM-1	65 (28 to 114)	0.04
Arteriolar sclerosis
FABP2	35 (16 to 56)	0.02
Acute tubular injury
SCF	−30 (–38 to –20)	<0.001
KIM-1	101 (52 to 166)	<0.001
TGF-α	27 (14 to 42)	0.004
PLXDC1	−16 (–22 to –9)	0.01
PTX3	33 (15 to 54)	0.04

Reference is none/mild lesion severity for acute tubular injury and arteriolar sclerosis. Reference is absence of lesion for mesangial expansion. Reference is 0%–10% of cortical volume or glomeruli affected for inflammation in the fibrosed and nonfibrosed interstitium and glomerular inflammation. Reference is 0%–25% of cortical volume or glomeruli affected for glomerular sclerosis and interstitial fibrosis/tubular atrophy. Endocapillary glomerular inflammation, extracapillary cellular crescents, focal glomerular necrosis, and fibrocellular crescents were combined into a single dichotomous variable named “glomerular inflammation.” Shown are significant associations after applying Bonferroni correction. Abbreviations and definitions of all proteins are provided in Supplemental Table 5.

^a Percentage differences were derived from linear regression models using plasma protein biomarkers as the dependent variable and each histopathologic lesion as the independent variable. Models were further adjusted for age, sex, race, and eGFR (modeled continuously).

Plasma Protein Biomarkers Associated with Kidney Disease Progression

In total, 170 individuals suffered kidney disease progression during a median follow-up of 52 months. Higher levels of 30 plasma proteins were independently associated with greater risk of kidney disease progression (Figure 2A, Supplemental Table 2). In multivariable-adjusted models, higher levels of the following biomarkers were associated with greater risks of kidney disease progression (top five findings in the order of level of significance): placental growth factor (PGF; hazard ratio [HR] per doubling, 5.4; 95% confidence interval [95% CI], 3.4 to 8.7), BMP and activin membrane-bound inhibitor (BAMBI; HR, 3.0; 95% CI, 2.1 to 4.2), TNF-receptor superfamily-11A (TNFRSF-11A), TNF-related apoptosis-inducing ligand-R2 (TRAIL-R2/TNFRSF-10B), and C-X3-C motif chemokine ligand 1. Among all 30 proteins, several were ILs and IL receptors (IL-4RA, IL-10RB, IL-15RA, and IL-16), hemostatic markers (thrombomodulin, protease-activated receptor-1, and tissue factor), and members of the TNF superfamily (TNFRSF-9, TNFRSF-10A, TRAIL-R2/TNFRSF-10B, and TNFRSF-11A). Cytokine signaling and activation of the clotting cascade were among the top-ranked pathways associated with kidney disease progression (Supplemental Table 3).

Figure 2.

Circulating plasma proteins associated with kidney disease progression and death. (A) Kidney disease progression, (B) death. Results are derived from proportional hazards models adjusted for eGFR, race, age, sex, log(proteinuria), and primary clinicopathologic diagnosis. Horizontal dotted lines show Bonferroni-adjusted significance thresholds, and vertical dotted lines mark the hazard ratio of one.

Plasma Protein Biomarkers Associated with Death

In total, 82 participants died during a median follow-up of 59 months. Higher levels of 30 proteins and lower levels of five proteins were independently associated with greater risk of death (Figure 2B, Supplemental Table 4). The top five proteins positively associated with greater risk of death according to level of significance were TRAIL-R2/TNFRSF-10B (HR per doubling, 2.9; 95% CI, 2.0 to 4.0), CUB domain-containing protein 1 (HR, 2.4; 95% CI, 1.8 to 3.3), TNFRSF-10A, programmed cell death-ligand 1 (CD274), and brain natriuretic peptide. The five plasma proteins inversely associated with greater risk of death were SCF, desmoglein-4, carbonic anhydrase 14, plexin domain-containing protein, and disintegrin and metalloproteinase with thrombospondin motifs-13 (ADAMTS13). Among all proteins associated with death, there were several members of the TNF superfamily (TNFRSF-10A, TNFRSF-10B/TRAIL-R2, and TNFRSF-11A) as well as cytokines and cytokine receptors involved in Th2 inflammation (IL-4RA and IL-6), NK cell regulation (IL-18R1, IL-15RA, and IL-27), and neutrophil recruitment (C-X-C motif chemokine 8/IL-8). Cytokine and IL signaling pathways were among the top-ranked pathways associated with death (Supplemental Table 3).

Discussion

This study provides an assessment of a subset of the plasma proteome in a cohort of individuals with biopsy-confirmed kidney diseases. We evaluated associations of 225 plasma biomarkers with histopathologic findings in native kidney biopsies and subsequent risks of kidney disease progression and death and identified several new promising biomarker candidates. Pathways enriched in the plasma proteome of patients at risk of disease progression may point toward key mechanisms involved in disease pathogenesis, including inflammation, extracellular matrix remodeling, disturbances of hemostasis, cell growth, and apoptosis.

Previous large-scale proteomics studies have primarily focused on evaluation of the urine CKD proteome (15–19). Fewer studies have utilized proteomics to interrogate plasma in patients with CKD (9,20–22). Some of the individual proteins that we investigated have previously been shown to be associated with histopathologic lesions and adverse clinical outcomes in individuals with kidney disease. Among those are plasma KIM-1 (1,23) and fibroblast growth factor-23 (4). Many others are new findings that represent areas for future investigation.

In line with results from previous studies, KIM-1 was the biomarker most strongly associated with more severe ATI, and it was also associated with a higher risk of kidney disease progression, which may provide additional evidence for the importance of kidney tubular injury in CKD progression (1,2,23,24). VSIG2, a protein of unknown function, was the top-performing marker for IFTA, and higher VSIG2 levels associated with a higher risk of kidney disease progression. VSIG2 has been associated with incident heart failure (25) and prevalent diabetes in the general population (26), but there are no studies, to our knowledge, that have linked VSIG2 to kidney diseases. Although we are unable to demonstrate a causal association, these results could point toward a potential role of VSIG2 in the progression of kidney fibrosis. Several inflammatory markers were associated with histopathologic lesions. EN-RAGE, a proinflammatory marker that leads to IL-1β and TNF-α release, had the strongest association with glomerular inflammation. In rodent models of CKD, EN-RAGE was reported to play a key role in the pathogenesis of vascular calcification through modulation of Pi cotransporter expression (27). This finding is supported by previous studies in humans that described associations between elevated plasma EN-RAGE levels and atherosclerosis as well as cardiovascular and all-cause mortality in patients on dialysis (28–30). Our study adds that EN-RAGE not only may be crucial for sustaining chronic systemic inflammation in individuals with kidney disease but also may be an important marker specific for glomerular inflammation. The single marker that remained significantly associated with vascular pathology after multivariable adjustment was FABP2, a protein expressed in epithelial cells of the small intestines and involved in the reabsorption and distribution of long-chain fatty acids (31). Polymorphisms in the FABP2 gene have been linked to insulin resistance, dyslipidemia, and cardiovascular diseases (31,32). We are not aware of any prior studies that have evaluated plasma FABP2 in kidney disease. Both EN-RAGE and FABP2 were not associated with adverse clinical events in this study. If the distinct signal for histopathologic lesions we observed can be confirmed in additional studies, these markers may be able to serve as specific tools to assess for vascular or glomerular inflammatory lesions.

A previous plasma proteomics study by Carlsson et al. (20) of individuals from the general population identified 20 plasma proteins that were associated with incident CKD. Although the associations in the study by Carlsson et al. (20) lost statistical significance after adjustment for baseline eGFR, seven of these proteins (TRAIL-R2, PGF, CD40, protease-activated receptor-1, fibroblast growth factor-23, CSF-1, and thrombomodulin) were also associated with kidney disease progression in our study and remained statistically significant after multivariable adjustment. We observed a particularly strong association for PGF, a member of the vascular endothelial growth factor family that has previously been studied in the setting of preeclampsia, where lower circulating maternal PGF levels were found to be associated with the disease (33,34). A prior study in kidney disease demonstrated that higher PGF levels associated with lower risks of AKI and mortality in patients undergoing cardiac surgery (35), but another study identified that higher PGF levels associated with a higher risk of incident CKD (20). We observed a higher risk of kidney disease progression and more severe IFTA with higher PGF levels. It is possible that in patients with established CKD, PGF levels may not have a pathogenic role but rather, reflect a compensatory response to sustained angiogenic imbalance that is present in the setting of chronic injury and more severe kidney fibrosis. We also observed a strong association between higher levels of BAMBI and greater risk of kidney disease progression. BAMBI serves as a regulator of angiogenesis and endothelial homeostasis through modulation of alternative TGF-β signaling pathways, such as extracellular signal–related kinase 1/2 and Smad1/5 (36,37). Diabetic BAMBI^−/− mice develop more severe albuminuria with increased activation of alternative TGF-β pathways compared with wild-type animals (36). In this study, we observed an association between higher, not lower, BAMBI levels and greater risks of kidney disease progression. Because our cohort primarily included patients with established CKD, these findings may point toward an important role of BAMBI in regulating antifibrotic responses, whereas its elevation could be due to the active and ongoing fibrotic processes in CKD. This, however, remains to be tested in additional cohort studies of CKD.

The top-performing biomarker associated with higher risks of death was TRAIL-R2, a key mediator of apoptosis, which was also associated with kidney disease progression. TRAIL-R2 is a death domain-containing receptor that is capable of transmitting apoptotic signals in response to TRAIL binding (38). TRAIL, a member of the TNF gene superfamily, has been shown to associate with albuminuria in diabetes and plays an important role in the progression of diabetic nephropathy by inducing kidney cell death (20,38).

We found lower levels of five plasma proteins to be associated with a higher risk of death. Among those was SCF, a regulator of bone marrow–derived stem cell migration and survival, known to act on early hematopoietic stem cells and primitive hematopoietic progenitor cells. Findings from rodent models suggest that SCF is crucial to mediate tubular epithelial cell survival after hypoxic injury (39). Similar findings were reported in mice in which local injection of SCF improved cardiac function after myocardial infarction (40). In the Malmö Diet and Cancer study, lower circulating levels of SCF associated with significantly higher risks of stroke and heart failure as well as cardiovascular and all-cause mortality (41). We observed similar findings in this study and found lower SCF levels to be associated with a higher risk of death in patients with kidney diseases. Evaluation of inflammatory parameters in the Malmö Diet and Cancer study also revealed that C-reactive protein levels and white blood cell counts correlated inversely with SCF (41). In line with this, we showed that SCF was inversely associated with both ATI and interstitial inflammation, which could suggest that SCF may attenuate inflammation in patients with kidney disease or that, in turn, inflammation may function as a negative regulator of SCF. We also found that lower levels of ADAMTS13 were associated with a greater risk of death. ADAMTS13 is a vWf-cleaving protease, and deficiency of plasma ADAMTS13 activity causes thrombotic thrombocytopenic purpura. In a case-control study of patients with malignant hypertension, ADAMTS13 levels correlated inversely with creatinine and were significantly lower in patients with more severe thrombotic microangiopathy and hypertension (42). Although the functional relevance and putative prognostic value of these proteins need to be further explored, several biomarkers shared common biologic pathways with pathophysiologic relevance for kidney diseases, including inflammation and extracellular matrix remodeling, apoptosis, angiogenesis, and endothelial dysfunction.

Significant strengths of our study include the relatively large number of protein biomarkers included in the analyses. Adjudicated histopathologic scores on lesion severity and the prospective study design with long-term follow-up data allowed us to test associations not only with histopathology, but also with major adverse clinical events, including kidney disease progression and death. Our study has several limitations that warrant consideration as well. First, we were not able to take historical SCr values prior to biopsy into account and did not account for BP and therapy at baseline, which could alter levels of biomarkers or an individual’s risk of disease progression or death. Although almost all participants had proteinuria data, we did not convert UPCR to UACR for participants who only had measurements of UPCR, which should still provide similar risk estimation (43). Furthermore, although the majority of BKBC participants had two or more follow-up eGFR values, we cannot rule out that kidney disease progression was more likely to be detected in individuals with more eGFR values ordered by their clinicians. We were also not able to compare plasma protein measurements with protein measurements in the urine. Limitations of the proteomics assay include that protein values are obtained in relative quantification units (normalized protein expression values) rather than as absolute values, which may limit comparability between studies. As we selected three Olink proteomics panels on the basis of their potential relevance for kidney disease, the protein biomarkers included in the study necessarily represent processes related to cardiovascular disease, inflammation, and organ injury. Thus, our pathway analyses are limited by the narrow range and nature of the protein markers studied. Lastly, we were not able to replicate our findings in a validation cohort, which needs to be addressed in future studies.

In summary, we identified a number of protein biomarkers associated with histopathologic lesions on native kidney biopsy specimens and subsequent kidney disease progression and death. Several of our biomarker findings merit development of quantitative assays for further replication in other prospective cohort studies to develop markers for noninvasive diagnosis and optimal prognosis. Studies using techniques such as Mendelian randomization may identify which of our biomarker findings represent therapeutic targets for patients with specific kidney diseases. Our study demonstrates how findings from a cohort of individuals with available kidney pathology and prospective clinical follow-up can uncover relevant pathways and biomarker candidates for understanding, diagnosing, and providing prognosis of kidney diseases.

Disclosures

B.D. Humphreys reports consultancy agreements with Chinook Therapeutics, Janssen, and Pfizer; ownership interest in Chinook Therapeutics; receiving research funding from Chinook Therapeutics and Janssen; receiving honoraria from the American Society of Nephrology; patents and inventions with Evotec, AG; serving on editorial boards for American Journal of Physiology Renal Physiology, JCI Insight, Kidney International, and Seminars in Nephrology; serving as Vice President of the American Society of Clinical Investigation (ASCI) and as an associate editor for JASN; and serving as a scientific advisor or member of Chinook Therapeutics SAB, the National Institute of Diabetes and Digestive and Kidney Diseases Board of Scientific Advisors, and RegMed XB Regenerative Medicine Crossing Borders SAB. H.G. Rennke reports consultancy agreements with Ionis Pharmaceuticals Inc. and receiving honoraria from Wolters Kluwer for book royalties for Renal Pathophysiology: The Essentials, 5th Ed. I.M. Schmidt is supported by the American Philosophical Society Daland Fellowship in Clinical Investigation. A. Srivastava reports consultancy agreements with CVS Caremark and Tate & Latham (medicolegal consulting); receiving honoraria from AstraZeneca, Bayer, and Horizon Therapeutics PLC; and speakers bureau for AstraZeneca. A. Srivastava is also supported by National Institutes of Health grant K23DK120811, National Institute of Diabetes and Digestive and Kidney Diseases Kidney Precision Medicine Project Opportunity Pool grant U2CDK114886, and core resources from George M. O’Brien Kidney Research Center at Northwestern University (Northwestern University-GoKIDNEY) grant P30DK114857. I.E. Stillman reports serving as a board member of the Organization for Renal Care in Haiti, Torch Inc. (no compensation), and legal consulting. S.S. Waikar reports consultancy agreements with Allena, BioMarin, CVS, GSK, JNJ, Mallinckrodt, Mass Medical International, Metro Biotechnology, Oxidien, Pfizer, Regeneron, Roth Capital Partners, Sironax, Strataca/3ive, Venbio, and Wolters Kluewer; receiving research funding from Vertex; serving as a scientific advisor or member of Kantum (scientific advisory board); receiving personal fees from Barron and Budd (versus Fresenius), Bunch and James, Cerus, CVS, GE Health Care, GSK, the Harvard Clinical Research Institute (also known as Baim), JNJ, Kantum Pharma, Mallinckrodt, Mass Medical International, Pfizer, the Public Health Advocacy Institute, Roth Capital Partners, Strataca, Takeda, Venbio, and Wolters Kluewer; receiving grants and personal fees from Allena Pharmaceuticals; and serving as an expert witness for litigation related to GE product Omniscan, an expert witness for litigation related to the Fresenius product Granuflo, an expert witness for litigation involving cisplatin toxicity, an expert witness for litigation related to the Gilead product Tenofovir, and an expert witness for litigation related to DaVita laboratory testing. S.S. Waikar is also supported by National Institutes of Health grants UH3DK114915, U01DK085660, U01DK104308, R01DK103784, and R21DK119751. P.C. Wilson reports receiving research funding from NovoNordisk. M. Zhuo is supported by National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases T32 award DK007199. All remaining authors have nothing to disclose.

Funding

This study was supported by National Institute of Diabetes and Digestive and Kidney Diseases grant R01DK093574 (to S.S. Waikar). This work was conducted with support from Harvard Catalyst, the Harvard Clinical and Translational Science Center (National Institutes of Health, National Center for Advancing Translational Sciences award UL1TR001102), and financial contributions from Harvard University and its affiliated academic health care centers.

Supplemental Material

This article contains supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.09380721/-/DCSupplemental.

Supplemental Material. Olink assay and adjudication of histopathologic outcomes.

Supplemental Table 1. Histopathologic scoring system for light microscopy.

Supplemental Table 2. Associations of plasma biomarkers with kidney disease progression.

Supplemental Table 3. Top-ranked pathways for plasma biomarkers associated with adverse clinical outcomes.

Supplemental Table 4. Associations of plasma biomarkers with death.

Supplemental Table 5. Abbreviations and list of all plasma biomarkers included in the analyses.