See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323649695

SERUM UROMODULIN PREDICTS A DECLINE IN KIDNEY FUNCTION

INDEPENDENTLY FROM THE PRESENCE OF TYPE 2 DIABETES

Article  in  Journal of the American College of Cardiology · March 2018

DOI: 10.1016/S0735-1097(18)32381-7

CITATIONS READS

0 69

7 authors, including:

Andreas Leiherer Axel Muendlein

Vorarlberg Institute for Vascular Investigation VIVIT Vorarlberg Institute for Vascular Investigation VIVIT
174 PUBLICATIONS   548 CITATIONS    227 PUBLICATIONS   1,664 CITATIONS   

SEE PROFILE SEE PROFILE

Christoph Saely Eva Maria Brandtner

Vorarlberg Institute for Vascular Investigation VIVIT Vorarlberg Institute for Vascular Investigation VIVIT
449 PUBLICATIONS   2,437 CITATIONS    70 PUBLICATIONS   213 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Cell Based Therapy for Diabetes View project

project of the general use of razoxane as radiosensitizing agent View project

All content following this page was uploaded by Andreas Leiherer on 14 December 2018.

The user has requested enhancement of the downloaded file.

 Original Article
The value of uromodulin as a new serum marker to
predict decline in renal function
Andreas Leiherer a,c,d, Axel Muendlein a,c, Christoph H. Saely a,b,c, Eva M. Brandtner a
Kathrin Geiger a, Peter Fraunberger c,d, and Heinz Drexel a,b,c,e
Background: Uromodulin is the most abundant protein in
urine. Low uromodulin has been found associated with
diabetes as well as with chronic kidney disease (CKD).
Whether it also predicts a future decline in kidney function
is not known.
Methods: We evaluated the association between serum
uromodulin and kidney function in 529 patients and
performed a genome-wide association study. Clinical
parameters including renal function were determined at
baseline and reassessed at a 4-year follow-up visit.
Results: Patients’ serum uromodulin levels were highest
associated with a polymorphism in the UMOD coding region
and were significantly correlated with estimated glomerular
filtration rate (eGFR) (r ¼ 0.242, P < 0.001) and, in an inverse
way, with the albumin–creatinine ratio (r ¼ 0.120,
P ¼ 0.012). Serum uromodulin concentrations were
significantly lower in patients with CKD (eGFR < 60 ml/min
per 1.73 m2) than in patients with normal kidney function
(71.9  29.0 vs. 169.1  76.1 ng/ml, P < 0.001) and the
same applied to patients with albuminuria (148.7  72.2 vs.
167.9  77.6 ng/ml, P ¼ 0.008) or hypertension
(160.9  74.0 vs. 181.8  87.8 ng/ml, P ¼ 0.037).
Prospectively, patients who developed CKD during the
follow-up had significantly less uromodulin in serum than
those who did not (126.8  42.3 vs. 180.2  79.1 ng/ml,
P ¼ 0.003). Serum uromodulin concentration was inversely
associated with the development of CKD, even after
adjustment for patients age, sex, genotype of the identified
polymorphism, hypertension and diabetes status, and eGFR
(odds ratio ¼ 0.263, P ¼ 0.019), and it significantly increased
the performance of a prediction model for CKD (C-statistics
0.844 vs. 0.804, P ¼ 0.049).
Conclusion: The current study is the first evaluating the
value of uromodulin in serum as a novel predictive
biomarker for renal function and for the incidence of CKD.
Keywords: chronic kidney disease, coronary patients,
kidney disease, polymorphism, renal biomarker, serum
uromodulin, Tamm–Horsfall-protein, UMOD
Abbreviations: ACE, angiotensin-converting enzyme;
ACR, albumin–creatinine ratio; ASA, acetylsalicylic acid;
AT, angiotensin; AUC, area under the curve; CAD,
coronary artery disease; CKD, chronic kidney disease; CRP,
C-reactive protein; eGFR, estimated glomerular filtration
rate; FJHN, familial juvenile hyperuricemic nephropathy;
110 www.jhypertension.com
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
FU, follow-up; GWAS, genome-wide association study;
KDOQI, Kidney Disease Outcomes Quality Initiative; MCAR,
missing completely at random; MCMC, Markov Chain
Monte Carlo; OR, odds ratio; PMM, predictive mean
matching; ROC, receiver operating characteristic; SNP,
single-nucleotide polymorphism; T2DM, type 2 diabetes
mellitus; TAL, thick ascending limb
INTRODUCTION
P
atients with CKD have a greatly increased risk of
cardiovascular events and cardiovascular death and
it has already been demonstrated that the decline of
kidney function is also an independent marker for these
events [1]. Assessing the diabetic state and albuminuria is
routine for estimating the decline of renal function, but
novel biomarkers are needed to significantly improve its
prediction in high risk patients.
Uromodulin, also known as Tamm–Horsfall protein, has
been suggested as a new therapeutic target for blood
pressure (BP) and renal function [2]. A recent genome-wide
association study (GWAS) has shown a link between single-
nucleotide polymorphisms (SNPs) in the UMOD gene,
coding for uromodulin and estimated glomerular filtration
rate (eGFR) [3]. Nevertheless there are, to date, only few
studies with conflicting results, addressing the association
between the uromodulin protein level and kidney disease.
Although one study found a high uromodulin concentra-
tion to be linked to the onset of chronic kidney disease
(CKD) [4], others found no statistically significant relation-
ship [5] or an association between low uromodulin con-
centration and impaired kidney function [6–8].
Journal of Hypertension 2018, 36:110–118
a
Vorarlberg Institute for Vascular Investigation and Treatment (VIVIT), bDepartment of
Medicine and Cardiology, Academic Teaching Hospital Feldkirch, Feldkirch, Austria,
c
Private University of the Principality of Liechtenstein, Triesen, Liechtenstein, dMedical
Central Laboratories, Feldkirch, Austria and eDrexel University College of Medicine,
Philadelphia, Pennsylvania, USA
Correspondence to Heinz Drexel, Vorarlberg Institute for Vascular Investigation and
Treatment (VIVIT), Feldkirch, Austria; Department of Medicine and Cardiology,
Academic Teaching Hospital Feldkirch, Carinagasse 47, A-6807 Feldkirch, Austria.
Tel: +43 5522 303 2670; fax: +43 5522 303 7533; e-mail: vivit@lkhf.at
Received 24 February 2017 Revised 6 June 2017 Accepted 25 July 2017
J Hypertens 36:110–118 Copyright ß 2017 Wolters Kluwer Health, Inc. All rights
reserved.
DOI:10.1097/HJH.0000000000001527
Volume 36  Number 1  January 2018

Although these studies have measured uromodulin in
urinary samples, several recent studies assessed it in
plasma or serum. These studies clearly demonstrated a link
between low uromodulin concentration and the presence
of CKD [9–12].
The analysis of uromodulin from blood samples may
overcome several preanalytic drawbacks with the determi-
nation of uromodulin in human urine [13] and has been
estimated to outperform that from urine [12]. Moreover, the
correlation between eGFR and uromodulin has been
reported to be more prominent with serum measurements
[14]. However, longitudinal measurements, which have
been recommended for more information and which could
evaluate the value of serum concentration of uromodulin as
a predictive marker [11,15], have not been performed yet.
Thus, the current study was designed to assess serum
uromodulin, renal function, and further relevant clinical
parameters at baseline and after a 4-year follow-up to assess
the association of serum uromodulin with kidney function
and the value of uromodulin to predict a future decline of
kidney function.
PATIENTS AND METHODS
Study patients and laboratory analyses
From September 2005 to April 2008, we enrolled 529
consecutive white patients who were referred to elective
coronary angiography for the evaluation of established or
suspected stable coronary artery disease (CAD). Angiogra-
phy procedures and basic laboratory measurements
were performed as described previously [16]. The current
study has been approved by the Ethics Committee of the
University of Innsbruck and adheres to the Declaration of
Helsinki. Hypertension was defined according to the Seventh
Report of the Joint National Committee on Prevention,
Detection, Evaluation, and Treatment of High Blood Pres-
sure [17]. Renal function was assessed based on eGFR
and classified as CKD in patients with eGFR less than
60 ml/min per 1.73 m2, according to the guidelines of the
Renal Association [18], referring to the National Kidney
Foundation – Kidney Disease Outcomes Quality Initiative
(KDOQI) [19]. Uromodulin levels were determined by
using ELISA kits (BioVendor, Brno, Czech Republic; catalog
no. RD191163200R), specific for human uromodulin with
an interassay variation less than 6.5% from patients’ serum
samples. According to Kolmogorov–Smirnov and Shapiro–
Wilk tests, serum uromodulin concentration was not nor-
mally distributed (Fig. S1, http://links.lww.com/HJH/A835)
and tertiles for serum uromodulin ranged from 21.6 to
125.3 ng/ml, from 125.3 to 190.1 ng/ml, and from 190.2 to
634.1 ng/ml.
At the follow-up visit after 3.5  1.1 years, a basic labo-
ratory analysis was performed as described above. Renal
function was reassessed in 340 patients.
Cardio-metabo chip and genotyping
DNA was extracted from patient blood samples and SNPs in
cardiovascular and metabolism genes were characterized
by using the Illumina Cardio-Metabo Chip technology
(Illumina Inc, San Diego, California, USA) as described
previously [20]. Patient samples were excluded if they
Journal of Hypertension
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Serum uromodulin and chronic kidney disease
had genotyping failure for more than 5% of the sites or if
they were identified as duplicates. From a total of 196 725
SNPs, those were excluded with a minor allele frequency
less than 0.5%. In addition, SNPs were also excluded, if
they failed to be genotyped in more than 5% of the study
population or if they were not in Hardy–Weinberg equi-
librium among samples (critical P value ¼ 0.0001). After
frequency and genotyping pruning, there were 130 909
SNPs left. Genetic data were available for 488 patients.
With these settings, we did not observe significant
genomic inflation (l ¼ 1.03). Chromosomal position and
strength of linkage disequilibrium between SNPs is given
for pair-wise r2 according to HapMap download version 2
and the program Haploview version 4.2 (BROAD Institute,
Cambridge, Massachusetts, USA) [21].
Statistical analysis
Differences in baseline characteristics were tested for sta-
tistical significance with the chi-squared tests for categorical
and Jonckheere–Terpstra tests for continuous variables,
respectively. Correlation analyses were performed calculat-
ing nonparametric Spearman rank correlation coefficients.
For comparing continuous or categorical variables between
baseline and follow-up, we used Wilcoxon and McNemar
tests, respectively. Z-transformation was applied before
logistic regression analysis and all results are given as
mean  SD; P values less than 0.05 were considered signifi-
cant. Apart from one exception, all data were analyzed
according to complete-case analysis because of a negligible
percentage of missing data. For the albumin–creatinine
ratio (ACR), we had 11% missing data completely at random
(MCAR) according to Little’s MCAR test. For our multivariate
analysis, which comprised ACR as a possible cofounder, we
used the Markov Chain Monte Carlo method with predictive
mean matching as multiple imputation method and applied
15 data sets to estimate the missing data for ACR.
To examine the potential utility of uromodulin as a
predictive biomarker [22], several binary logistic regression
models were fitted with the drop below an eGFR of 60 ml/
min per 1.73 m2 as the dependent variable. A basic model
was built comprising age, sex, BP, BMI, the type 2 diabetes
mellitus (T2DM) status, and the presence of CAD according
to coronary angiography as independent variables. This
model was compared with further models comprising
uromodulin or ACR as additional marker. Receiver operat-
ing characteristic (ROC) and the respective area under
the curve (AUC) were calculated by using the pROC
package for R applying the DeLong’s test as described
elsewhere [23].
A-priori power calculation for a correlation (two-tailed)
with a correlation coefficient of 0.2 and a sample size of 529
with an alpha-fault of 0.05 showed that the expected power
is 99.7%.
To account for multiple testing in the GWAS, Bonferroni
correction was used to adjust for the critical P value thresh-
old (P value ¼ 3.81e  7).
All statistical analyses were performed with SPSS 21.0,
SPSS Sample Power 3.0 (SPSS, Inc., Chicago, Illinois, USA),
PLINK version 1.07 [24]; http:// http://zzz.bwh.harvard.
edu/plink/download.shtml, and R statistical software ver-
sion 3.2.3 (http://www.r-project.org).
www.jhypertension.com 111
Leiherer et al.

TABLE 1. Patient characteristics according to tertiles of serum uromodulin

Total n 1st tertile 2nd tertile 3rd tertile P value
Serum uromodulin (ng/ml) 164.9  77.2 529 90.1  23.8 156.1  18.9 248.2  65.2 <0.001
Age (years) 65.3  11.1 529 67.9  11.4 64.8  11.2 63.1  10.2 <0.001
Male sex (%) 64.7 529 65.0 66.3 62.7 0.657
BMI (kg/m2) 28.0  4.4 529 28.5  4.8 27.8  4.4 27.7  4.1 0.246
Waist circumference (cm) 99.2  11.5 487 100.5  11.9 98.5  11.9 98.7  10.7 0.235
CRP (mg/dl) 0.44  0.73 529 0.47  0.58 0.44  0.68 0.39  0.89 0.006
Fibrinogen (mg/dl) 330  74 524 335  76 330  77 323  68 0.111
LDL cholesterol (mg/dl) 130  40 529 127  40 130  38 133  40 0.084
HDL cholesterol (mg/dl) 58  17 527 57  16 58  15 59  18 0.243
Total cholesterol (mg/dl) 197  43 527 194  43 196  42 201  45 0.096
Triglycerides (mg/dl) 132  72 527 132  70 133  71 130  74 0.560
eGFR (ml/min per 1.73 m2) 96.9  19.1 529 89.1  23.5 99.7  15.5 101.9  14.4 <0.001
ACR (mg/mg) 79.9  269.1 468 101.3  268.6 101.3  364.5 34.0  75.6 0.015
CKD (%) 33.6 529 47.5 29.1 24.3 <0.001
Hypertension (%) 80.9 529 85.9 79.4 77.4 0.043
T2DM (%) 27.6 529 36.7 24.6 21.5 0.001
CAD (%) 55.2 529 55.9 50.9 58.8 0.593
Current smoking (%) 17.4 529 14.1 17.7 20.3 0.123
Statin treatment (%) 51.4 529 48.6 49.7 55.9 0.167
ASA treatment (%) 68.8 529 65.5 68.0 72.9 0.136
Beta blocker treatment (%) 54.4 529 56.5 56.0 50.9 0.286
ACE inhibitor treatment (%) 29.5 529 33.9 27.4 27.1 0.162
AT-2 antagonist treatment (%) 12.7 529 13.0 15.4 9.6 0.338

Patient characteristics were shown in tertiles of serum uromodulin concentration. Data are given as means  SDs as indicated. P values are given for trend. ACE, angiotensin-converting
enzyme; ACR, albumin-to-creatinine ratio (mg albumin/mg creatinine); ASA, acetylsalicylic acid; AT-2, angiotensin 2; CAD, coronary artery disease defined by an angiographically
determined coronary artery stenosis with lumen narrowing at least 50%; CKD, chronic kidney disease defined by an eGFR smaller than 90 ml/min per 1.73 m2; CRP, C-reactive protein;
eGFR, estimated glomerular filtration rate; T2DM, type 2 diabetes mellitus as defined by American Diabetes Association.

RESULTS Genome-wide association with serum

Patient characteristics and subgroup analysis
Table 1 presents patient characteristics for the total study
population as well as in tertiles of serum uromodulin. The
mean  SD of serum uromodulin concentration was
164.9  77.2 ng/ml, with a median of 155.8 ng/ml (inter-
quartile range 109.4–204.3 ng/ml).
As a continuos variable, uromodulin was significantly
associated with age (r ¼ 0.208, <0.001), fasting glucose
(r ¼ 0.161, P < 0.001), and the C-reactive protein (CRP;
r ¼ 0.133, P ¼ 0.002). Older patients (>65 years) had a
significantly lower serum uromodulin concentration
than younger (65 years) patients (154.3  69.4 vs.
178.4  84.4 ng/ml, P ¼ 0.001) and the same applied
to patients with hypertension compared with those
without hypertension (160.9  74.0 vs. 181.8  87.8 ng/ml,
P ¼ 0.037).
uromodulin
Performing a GWAS with serum uromodulin as phenotype,
we revealed highest association for SNP rs13335818. This
SNP is in perfect linkage disequilibrium with rs12917707
(r2 ¼ 1.0), for which a previous meta-analysis by Olden
et al. has demonstrated the strongest association with urinary
uromodulin: the minor allele was associated with lower
urinary uromodulin but higher eGFR [3]. A similar result
was seen with our data for serum uromodulin (Table 2).
Results of the GWAS are presented in Supplemental part (Fig.
S2 and Table S1, http://links.lww.com/HJH/A835).
Association of serum uromodulin with
parameters of kidney function
Serum uromodulin did not correlate with urinary creatinine
(r ¼ 0.025, P ¼ 0.583), but correlated inversely with uri-
nary albumin (r ¼ 0.185, P < 0.001) and with the ACR

TABLE 2. Stratification according to rs13335818 genotype

rs13335818 N Uromodulin (ng/ml) eGFR (ml/min per 1.73 m2) r for uromodulin and eGFR
C/C 312 185.5  75.2 Ptrend < 0.001 96.0  19.9 Ptrend ¼ 0.406 0.283 P < 0.001
C/T 158 126.9  52.0 97.7  18.6 0.215 P ¼ 0.007
T/T 18 84.4  27.0 99.5  11.0 0.193 P ¼ 0.443
C/C 312 185.5  75.2 P < 0.001 96.0  19.9 P ¼ 0.418 0.283 P < 0.001
X/T 176 122.6  51.7 97.9  18.0 0.209 P ¼ 0.005

Polymorphism rs13335818 is the top single-nucleotide polymorphism identified in the genome-wide association study on serum uromodulin and in perfect linkage disequilibrium
(r2 ¼ 1.0) with rs12917707, identified as top single-nucleotide polymorphism in a previous genome-wide association study on urinary uromodulin [3]. Population was stratified according
to the respective rs13335818 genotype (additive and dominant model for the minor allele) and serum uromodulin, estimated glomerular filtration rate (eGFR), and the correlation (r)
between uromodulin and eGFR are given with the respective P values (Jonckheere–Terpstra test for trend and Mann–Whitney U test).

112 www.jhypertension.com Volume 36  Number 1  January 2018

Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.

 Serum uromodulin and chronic kidney disease

TABLE 3. Analysis of covariance for the association of estimated glomerular filtration rate with serum uromodulin
eGFR
Model Covariates F P value
1 – 33.0 <0.001
2 þ rs13335818 48.2 <0.001
3 þ age, sex 47.6 <0.001
4 þ BP, CRP, BMI 48.0 <0.001
5 þ ACR, T2DM, CAD 42.3 <0.001

Association of estimated glomerular filtration rate with uromodulin in serum samples is given as F test and P values for five different models. Model 1 represents a univariate analysis.
Model 2 was adjusted for rs13335818 genotypes; model 3 in addition comprises covariates age and sex; model 4 comprises all covariates of model 3 including hypertension status (BP),
CRP, and BMI; model 5 comprises all covariates of model 4 including ACR as well as T2DM-status and CAD-status. ACR, albumin–creatinine ratio; BP, blood pressure; CAD, coronary
artery disease; CRP, C-reactive protein; T2DM, type 2 diabetes mellitus.

(r ¼ 0.120, P ¼ 0.012). Patients with macroalbuminuria
(ACR > 300) had a lower serum uromodulin level
(145.0  81.5 ng/ml) than those with microalbuminuria
(149.8  69.8 ng/ml) and patients with normalbuminuria
had the highest concentration of uromodulin in serum
(167.9  77.6 ng/ml, Ptrend ¼ 0.006).
There were significant correlations between serum uro-
modulin and serum creatinine (r ¼ 0.288, P < 0.001) as
well as between serum uromodulin and serum urea
(r ¼ 0.314, P < 0.001). Serum and urinary creatinine were
correlated as well (r ¼ 0.166, P < 0.001).
We found a positive correlation between eGFR and
serum uromodulin concentration for the whole study
(r ¼ 0.242, P < 0.001). After stratification according to ge-
notype of rs13335818, correlation was significant in patients
with two major alleles (r ¼ 0.283, P < 0.001) as well as in
patients with at least one minor allele (r ¼ 0.209, P ¼ 0.005).
The minor allele appeared to be slightly linked with higher
eGFR, resulting in a shift of the correlation between uro-
modulin and eGFR (Table 2, Supplementary Fig. S3, http://
links.lww.com/HJH/A835).
In a univariate linear regression model, we calculated a
regression coefficient B of 0.060 together with a change
(beta) of 0.243 (P < 0.001) with an adjusted R2 of 5.5%,
and the statistic effect size ( f2) of uromodulin on eGFR was
0.25. In line with univariate results, analysis of covariance
adjusting for age, sex, genotypes of rs13335818, and hyper-
tension and diabetes status revealed that serum uromodulin
was significantly and independently associated with eGFR
(F ¼ 43.5; P < 0.001). Further adjustment models are sum-
marized in Table 3.
In addition, patients with CKD (eGFR < 60 ml/min per
1.73 m2) displayed significantly lower uromodulin levels
than those with higher eGFR (70.7  28.4 vs. 169.0  76.0,
P < 0.001). Multivariate logistic regression analysis demon-
strated that increasing serum uromodulin concentration
was strongly and significantly associated with a decreasing
prevalence of CKD (Fig. 1).
We further observed a significant decline in serum
uromodulin from patients with normal renal filtration rates
(eGFR  90 ml/min per 1.73 m2) over CKD stage 2 (eGFR
60–89 ml/min/1.73 m2), stage 3A (eGFR 45–59 ml/min
per 1.73 m2), stage 3B (eGFR 30–44 ml/min per 1.73 m2)
to stage 4 (eGFR 15–29 ml/min per 1.73 m2, Ptrend < 0.001).
The difference in serum uromodulin between moderately
reduced kidney function (CKD stage 3A) and mildly
reduced kidney function (CKD stage 2) was significant
(85.4  34.8 vs. 152.3  67.3 ng/ml, P ¼ 0.004). In addition,
patients with mildly reduced kidney function also had
significantly lower uromodulin levels than those with
normal renal filtration rates (152.3  67.3 vs. 176.4 
78.6 ng/ml, P ¼ 0.001; Fig. 2).
In multivariate logistic regression analysis, serum uro-
modulin proved to be a significant predictor for an eGFR
< 90 ml/min per 1.73 m2, even after adjustment for age, sex,
genotypes of rs13335818, and hypertension and diabetes
status [odds ratio (OR) ¼ 0.489 (95% confidence interval
(CI) 0.342–0.701), P < 0.001, further adjustment models are
given in Fig. S4, http://links.lww.com/HJH/A835].
Prospective data
Associations between uromodulin and kidney parameters
at the 4-year follow-up were comparable with those deter-
mined at baseline (Table 4, Table S2, Fig. S5, http://links.
lww.com/HJH/A835).
In patients with normal eGFR (60 ml/min per 1.73 m2)
at baseline, uromodulin was significantly and inversely
correlated with the decline of eGFR during the follow-up
Model 1
0.028 [0.008-0.102] P < 0.001
Model 2
0.012 [0.003-0.057] P < 0.001
Model 3
0.011 [0.002-0.063] P < 0.001
Model 4
0.012 [0.002-0.067] P < 0.001
Model 5
0.011 [0.004-0.028] P < 0.001
2-9 2-8 2-7 2-6 2-5 2-4 2-3 2-2 2-1 1 2
FIGURE 1 Association between serum uromodulin concentration and the preva-
lence of chronic kidney disease. The Forest plot represents the odds ratios with
95% confidence interval for the association between the prevalence of chronic
kidney disease (estimated glomerular filtration rate <60 ml/min/1.73 m2) and
serum uromodulin concentration in our study population with stepwise adjustment.
Model 1 represents univariate analysis. Adjustment model 2 includes the
rs13335818 genotypes. Adjustment model 3 includes the parameter included in
model 2 and in addition, the covariates age and sex. Adjustment model 4 includes
the parameters included in model 3 and in addition, hypertension, C-reactive pro-
tein, and BMI. Adjustment model 5 includes the parameters included in model 4 and
in addition, the covariate albumin–creatinine ratio as well as presence of type 2
diabetes mellitus and coronary artery disease.

Journal of Hypertension www.jhypertension.com 113

Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Leiherer et al.

CKD (eGFR > 60 ml/min per 1.73 m2) who developed CKD
*** during follow-up with those who did not develop CKD, and
**
200 we observed a significant difference (126.8  42.3 vs.
**
n.s. 180.2  79.1 ng/ml; P ¼ 0.003). On separating patients into
n.s. those with high and low uromodulin in serum, we found
176.4 that 80.0% of all patients who developed CKD during
follow-up were in the low uromodulin group (P ¼ 0.003)
152.3 and similar results are seen on using uromodulin tertiles
150
(53.3, 40.0, 6.7%, Ptrend ¼ 0.003).
Uromodulin (ng/ml)

Finally, applying logistic regression analysis revealed a

100
85.4
63.7
50
0 dence of CKD in patients with the homozygous major allele
≥ 90 60-89 45-59 30-44
eGFR (ml/min per 1.73 m2)
FIGURE 2 Comparison of serum uromodulin in stages of chronic kidney disease.
The means of serum uromodulin with 95% confidence interval (error bars) are
depicted for chronic kidney disease stages according to KDOQI [41]. A total of
351 patients were classified normal with estimated glomerular filtration rate at
least 90 ml/min per 1.73 m2, 156 with mildly reduced but still normal estimated
glomerular filtration rate 60–89 ml/min per 1.73 m2 (stage 2), seven stage 3A
(estimated glomerular filtration rate 45–59 ml/min per 1.73 m2), 12 stage 3B
(estimated glomerular filtration rate 30–44 ml/min per 1.73 m2), and three stage
4 chronic kidney disease (estimated glomerular filtration rate 15–29 ml/min per
1.73 m2). P < 0.001, P < 0.005.
time (r ¼ 0.130, P ¼ 0.018), and also significantly correlat-
ed with the change of ACR during the follow-up time
(r ¼ 0.150, P ¼ 0.010).
With respect to the incidence of CKD, we also compared
the uromodulin baseline concentration of patients without
significant inverse association between serum uromodulin
concentration as a continuous variable and the develop-
ment of CKD during follow-up [eGFR < 60 ml/min per
1.73 m2; OR ¼ 0.353 (95% CI 0.165–0.755), P ¼ 0.007] in
patients with normal eGFR at baseline. This association
remained significant, even after adjustment for patients’
age, sex, genotypes of rs13335818, hypertension and dia-
64.4 betes status, as well as eGFR at baseline [OR ¼ 0.263 (95%
CI 0.087–0.799), P ¼ 0.019; further adjustment models are
given in Fig. 3].
Patients’ genotype was not significantly associated with
the development of CKD [OR ¼ 0.311 (95% CI 0.073–1.322),
P ¼ 0.114]. Uromodulin concentration, if stratified with re-
spect to the genotype, was only associated with the inci-
15-29 [OR ¼ 0.131 (95% CI 0.043–0.396), P < 0.001] but not in
patients with at least one minor allele [OR ¼ 1.054 (95% CI
0.149–7.434), P ¼ 0.958].
The value of uromodulin as predictive
biomarker
We finally examined the utility of uromodulin to predict
renal decline according to eGFR. Table 5 and Supplemen-
tary Fig. S6, http://links.lww.com/HJH/A835 give an over-
view about different prediction models. The inclusion of
uromodulin to the basic prediction model led to a signifi-
cant gain with respect to the AUC for the ROC. Moreover,
the use of uromodulin as a one-variable model resulted in
an AUC that was not significantly inferior to the basic
model. In contrast, the inclusion of ACR instead of uromo-
dulin did not significantly increase the performance of the
basic model (Table 5).

TABLE 4. Serum uromodulin concentration in subgroups with respect to estimated glomerular filtration rate thresholds and albuminuria
at baseline and follow-up
Baseline Follow-up
2
eGFR (ml/min/1.73 m ) Uromodulin (ng/ml) P value Uromodulin (ng/ml) P value
<60 70.7  28.4 n ¼ 22 <0.001 106.8  46.2 n ¼ 22 <0.001
60 169.0  76.0 n ¼ 507 179.7  79.4 n ¼ 318
<90 142.2  69.2 n ¼ 178 <0.001 163.4  72.8 n ¼ 141 0.018
90 176.4  78.6 n ¼ 351 183.3  83.5 n ¼ 199

Baseline Follow-up
ACR (mg albumin/mg creatinine) Uromodulin (ng/ml) Ptrend value Uromodulin (ng/ml) Ptrend value
>300 145.0  81.5 n ¼ 27 0.006 146.3  82.5 n ¼ 18 0.006
30–300 149.8  69.8 n ¼ 97 158.4  72.8 n ¼ 52
<30 167.9  77.6 n ¼ 344 180.0  80.6 n ¼ 263

Serum uromodulin concentration is given for patients according to thresholds of eGFR as indicated and according to the ACR calculated as mg urinary albumin to mg urinary creatinine,
defining macroalbuminuria (ACR > 300), microalbuminuria (ACR 30–300), and normalbuminuria (ACR < 30), with the respective P values. ACR, albumin–creatinine ratio; eGFR,
estimated glomerular filtration rate

114 www.jhypertension.com Volume 36  Number 1  January 2018

Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.

 Serum uromodulin and chronic kidney disease

Most of our 529 patients had normal or only moderately

Model 1
0.353 [0.165-0.755] P = 0.007
Model 2
0.179 [0.069-0.467] P < 0.001
Model 3
0.227 [0.081-0.637] P = 0.005
Model 4
0.220 [0.074-0.653] P = 0.006
Model 5
0.164 [0.043-0.620] P = 0.008
2 -6 2 -5 2 -4 2 -3 2 -2 2 -1 1 2
FIGURE 3 Association between serum uromodulin concentration and the incidence
of chronic kidney disease. The Forest plots depicts the odds ratios and 95% confi-
dence interval derived from binary logistic regression analysis to predict the incidence
of chronic kidney disease (estimated glomerular filtration rate <60 ml/min per
1.73 m2) by using serum uromodulin concentration as a continuous variable with
stepwise adjustment. Adjustment model 1 represents univariate analysis. Adjustment
model 2 includes the rs13335818 genotypes. Adjustment model 3 includes the
covariates genotype, age, and sex. Adjustment model 4 includes the parameters
included in model 3 and in addition, BMI, hypertension, and C-reactive protein.
Adjustment model 5 includes the parameters included in model 4 and in addition,
the albumin–creatinine ratio and the estimated glomerular filtration rate at baseline,
the type 2 diabetes mellitus status, and the coronary artery disease status.
DISCUSSION
Serum uromodulin is associated with kidney
function
In our study, uromodulin serum levels were inversely
correlated with creatinine and urea in serum as well as
with urinary albumin. We also observed a significant asso-
ciation between uromodulin and CKD, albuminuria, and
hypertension. This link between impaired kidney function
and uromodulin concentration corroborates previous find-
ings by Steubl et al. These authors compared 71 patients
without CKD with 355 patients at various stages of CKD,
allowing identification of early stages of CKD [12]; similar
results were also reported by Fedak et al. [11]. In addition,
our data are also in accordance with those from Risch et al.
[9] for 289 subjectively healthy elderly (60 years) patients
without diabetes, reporting an association between lower
serum and plasma uromodulin levels and decreased kidney
function [9] and also with earlier studies of limited sample
size, mentioned therein [25,26].
reduced kidney function and only a minority exhibited a
more severe decline of kidney function. We were thus able
to clearly discriminate between normal and mildly reduced
as well as between mildly reduced and moderately reduced
renal filtration.
In contrast to Risch et al. [9], almost a third of our
participants had diabetes and although uromodulin is also
known to be linked to diabetes [27], we did not see a
significant impact of diabetes on the power of uromodulin
to predict renal function in our multivariate analyses. Fur-
thermore, our data on serum uromodulin are also in line
with prior studies, linking decreased urinary uromodulin
excretion with a low glomerular filtration rate [28,29].
We also observed a significant and inverse correlation
between uromodulin on the one hand and the inflamma-
tory marker CRP on the other. These data are again in line
with previous reports demonstrating increased neutrophil
infiltration [30], more tubular necrosis, inflammation, and
renal impairment in uromodulin knockout mice than in
wild-type mice [31], as well as a negative correlation be-
tween the gene expression of inflammatory cytokines and
uromodulin for hypertension patients [32], arguing for the
renoprotective properties of uromodulin [14,33].
Serum uromodulin is associated with UMOD
polymorphisms
In our GWAS with serum uromodulin concentration, we
have identified rs13335818 as a top-ranked polymorphism.
It is located in the UMOD gene, causing a synonymous
codon change and is in perfect linkage disequilibrium with
SNP rs12917707, previously identified as top polymorphism
in a genome-wide association with uromodulin in urine as
phenotype [3]. This polymorphism is located in the up-
stream UMOD gene region and its minor allele is known to
be associated with lower urinary [3] and serum uromodulin
concentration [34], but astonishingly with higher eGFR
[3,34]. Similarly, we also found lower serum uromodulin
concentration for the minor allele of the linked polymor-
phism rs13335818, although the association with higher
eGFR was only by trend, failing significance. Thus, our
regression models were adjusted for rs13335818, strength-
ening the association of uromodulin with kidney function.
Serum uromodulin is a predictive marker for
future renal decline
Our study for the first time shows that low serum uromo-
dulin concentration is not only an indicator for the preva-
lence of kidney disease but also a predictor for future

TABLE 5. Comparison of the area under the curve of prediction models for decline of glomerular filtration
Model AUC (95% CI) P value for comparison with basic model Z for DeLong’s test
Basic 0.804 (0.668–0.939) – –
Basic þ ACR 0.832 (0.711–0.953) 0.307 1.023
Basic þ uromodulin 0.844 (0.719–0.969) 0.049 1.960
Uromodulin 0.753 (0.641–0.866) 0.429 0.790

The first model comprises age, sex, BMI, blood pressure, the diabetes mellitus type 2 status, and the presence of coronary artery disease, and was determined ‘basic’. The basic model
(basic) was compared with models additionally comprising, the ACR (basic þ ACR), or uromodulin (basic þ uromodulin). A single variable model consisting solely of uromodulin was
determined ‘uromodulin’. Models were built as binary logistic regression model. The AUC for the receiver operator characteristic is given for the decline of estimated glomerular
filtration rate (<60 ml/min/1.73 m2). AUC was compared according to DeLong’s test with the respective P values given for comparison with the basic model (basic). ACR, albumin–
creatinine ratio; AUC, area under the curve; CI, confidence interval.

Journal of Hypertension www.jhypertension.com 115

Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Leiherer et al.
decline of renal function. The risk for developing CKD was
inversely associated with serum uromodulin concentration,
independent from other risk factors including patients’ BP,
diabetes, CAD, as well as kidney markers ACR and eGFR. A
stratification of the study population according to the
UMOD genotype (rs13335818) revealed that uromodulin
is only a significant predictor for kidney decline in patients
homozygous for the major allele but not in patients with at
least one minor allele. These data imply that uromodulin
measurement may become most meaningful if it is inter-
preted on the respective genetic background. Whether the
mentioned polymorphism in UMOD confers to a worse
codon usage-based translational rate or cis-acting effects
modulating the transcription, as supposed for rs12917707
[3], posttranscriptional regulation, or RNA stability or if
rs13335818 is genetically linked to a different more severe
variant is unknown and further studies are necessary to
reveal the mechanistic background.
Of note, studies to predict the incidence of CKD based
on uromodulin from urinary samples in the past have rarely
been conducted and delivered contradictory results. In a
previous case–control study with 200 patients by Köttgen
et al. higher uromodulin concentration in urine was linked
to the onset of CKD [4], whereas two more recent studies
analyzing urine from 344 IgA nephropathy [35] and 49 CKD
patients [36] described an association between low urinary
excretion of uromodulin and future decline of kidney
function. These discrepancies may be due to different
detection methods used in the studies (gel electrophoresis,
immunoarray, spectrometry-based techniques), polymor-
phisms in the uromodulin coding gene region [4], and the
general constraints of urine sample measurements, which
finally led to the use of blood samples for measuring of
uromodulin in recent studies [9–14,37].
Taken together, the vast majority of previous and recent
studies on urinary uromodulin as well as recent studies on
uromodulin from patients’ blood clearly support the link
between low uromodulin and impaired kidney function.
Thus, our prospective data demonstrates that serum uromo-
dulin is also inversely associated with the decline of renal
function and the incidence of CKD respectively fits very well
into the current picture. Moreover, serum uromodulin has
recently been demonstrated to predict mortality, indepen-
dent from eGFR at baseline [34,38]. As a decrease of kidney
function is a biomarker for the risk of mortality and cardio-
vascular risk [1], the prediction of future renal decline facili-
tated by a single assessment of serum uromodulin is valuable
and, more importantly, gives early information about
patients’ long-term health status and survival.
The value of uromodulin in a prediction model
We calculated ROC curves to compare different risk pre-
diction models including only age, sex, BMI, BP, and the
T2DM and CAD-status, traditional risk factors for CKD, and
either uromodulin or ACR. The model including uromodu-
lin had the best performance represented by the highest
AUC. In addition, the difference in performance between
this model and the basic model without uromodulin was
significant. Although the inclusion of ACR increased the
performance of the basic model as well, this gain in perfor-
mance was not significant. On the other hand, the
116 www.jhypertension.com
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
difference in performance between the basic model com-
prising age, sex, BMI, BP, and the T2DM and CAD-status,
and the one-variable model with uromodulin as the
sole variable, was not significant. This further emphasizes
the predictive power of uromodulin. According to our
study data, uromodulin is a valuable predictor for future
renal decline and the significant increase in C-statistics
suggests that it provides even better predictive performance
than ACR.
Nevertheless, we also want to mention that there is a
large overlap between AUC confidence intervals, which
may raise the issue of biological significance and that results
from C-statistics, although statistically significant, give only
limited information about the biological situation.
Study strengths and limitations
The current study has limitations. First, we had mainly
patients with normal or mildly reduced kidney function
and only a limited number of patients with a more severe
reduction. On the other hand, this fact strengthens the study
as the presence of severe renal dysfunction and the predic-
tion of (further) renal impairment has limited value as
patients are most likely already under treatment. Second,
our study patients were undergoing coronary angiography
for the evaluation of CAD and are therefore a selected
group and thus, do not necessarily reflect the general
population. However, the population we chose to study
is clinically very important and is at a very high risk of
cardiovascular events. Important strengths of our work are
the long follow-up and the well characterization of patients.
Pathophysiological considerations and open
questions
There are still some open questions. Since decades, it has
been thought that a decreased urinary uromodulin excre-
tion is somehow linked with a low glomerular filtration rate
[28,39] and GWAS has already identified common variants
of the uromodulin gene (UMOD) to be associated with
kidney function [3,40]. Nevertheless, we don’t know until
now the detailed function or the regulation of uromodulin
nor the source and detailed transport in case of serum
uromodulin. In that context, the relation between uromo-
dulin in urine and blood is a further unsolved issue.
Moreover, even if uromodulin in serum has now been
demonstrated to predict kidney dysfunction very early and
sensitively, it should be kept in mind that uromodulin is
thought to be a marker of renal mass and the number of
functional tubular cells, whereas eGFR is primarily a marker
for filtration with limited value to present kidney function as
a whole [7,9,12,26,41].
Recently, it is has been proposed that it is the absolute
uromodulin excretion that decreases, whereas the amount
of uromodulin secreted by each single functioning nephron
unit increases in renal disease [15]. This is in line with data
from diabetes-induced rats [42,43] and diabetic humans
[44–46], demonstrating increased uromodulin excretion
at a time when kidney function is still unimpaired. On
the contrary, in our study population as well as in most
others, the time when patients’ kidney function is starting to
get worse, if at all, is obscure.
Volume 36  Number 1  January 2018

Uromodulin is expressed primarily in the thick ascend-
ing limb and even if expression elsewhere [47,48] has been
regarded negligible [49], in terms of serum concentrations
of uromodulin, laying beyond that of urine by factor of
about 1000, as suggested previously [33,41], it might be
of relevance.
In conclusion, the current study is currently the largest
investigation analyzing serum uromodulin with respect to
kidney function and it demonstrates for the first time that
low serum uromodulin concentration is linked with the
incidence of CKD. We thus hope that our data contribute to
enlarge the pathophysiological picture of uromodulin.
Apart from that, we also recommend considering the inclu-
sion of uromodulin for standardized testing next to eGFR
and ACR, as it is not only a marker for renal mass but is also
proved here to be an independent, sensitive, and predictive
biomarker for renal function.
ACKNOWLEDGEMENTS
A.L. received uromodulin ELISA Kits from BioVendor
(Brno, Czech Republic), apart from that, there are no
competing financial interests. We thank the Medical Central
Laboratories at the Academic Teaching Hospital Feldkirch
(Feldkirch, Austria), the Austrian National Bank, and the
Land Vorarlberg for providing us with generous research
grants. We also thank Dr Jochen Hauer from BioVendor
(Brno, Czech Republic) for dynamic support.
Conflicts of interest
There are no conflicts of interest.
REFERENCES
1. Rein P, Saely CH, Vonbank A, Boehnel C, Drexel H. Usefulness of serial
decline of kidney function to predict mortality and cardiovascular
events in patients undergoing coronary angiography. Am J Cardiol
2014; 113:215–221.
2. Trudu M, Janas S, Lanzani C, Debaix H, Schaeffer C, Ikehata M, et al.
Common noncoding UMOD gene variants induce salt-sensitive hyper-
tension and kidney damage by increasing uromodulin expression. Nat
Med 2013; 19:1655–1660.
3. Olden M, Corre T, Hayward C, Toniolo D, Ulivi S, Gasparini P, et al.
Common variants in UMOD associate with urinary uromodulin levels: a
meta-analysis. J Am Soc Nephrol 2014; 25:1869–1882.
4. Köttgen A, Hwang SJ, Larson MG, Van Eyk JE, Fu Q, Benjamin EJ, et al.
Uromodulin levels associate with a common UMOD variant and risk for
incident CKD. J Am Soc Nephrol 2010; 21:337–344.
5. Shlipak MG, Li Y, Fox C, Grunfeld C, Whooley M. Uromodulin con-
centrations are not associated with incident CKD among persons with
coronary artery disease. BMC Nephrol 2011; 12:2.
6. Chakraborty J, Below AA, Solaiman D. Tamm–Horsfall protein
in patients with kidney damage and diabetes. Urol Res 2004; 32:
79–83.
7. Garimella PS, Biggs ML, Katz R, Ix JH, Bennett MR, Devarajan P, et al.
Urinary uromodulin, kidney function, and cardiovascular disease in
elderly adults. Kidney Int 2015; 88:1126–1134.
8. Lynn KL, Marshall RD. Excretion of Tamm–Horsfall glycoprotein in
renal disease. Clin Nephrol 1984; 22:253–257.
9. Risch L, Lhotta K, Meier D, Medina-Escobar P, Nydegger UE, Risch M.
The serum uromodulin level is associated with kidney function. Clin
Chem Lab Med 2014; 52:1755–1761.
10. Satanovskij R, Bader A, Block M, Herbst V, Schlumberger W, Haack T,
et al. A new missense mutation in UMOD gene leads to severely
reduced serum uromodulin concentrations – a tool for the
diagnosis of uromodulin-associated kidney disease. Clin Biochem
2017; 50:155–158.
Journal of Hypertension
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
Serum uromodulin and chronic kidney disease
11. Fedak D, Kuzniewski M, Fugiel A, Wieczorek-Surdacka E, Przepiór-
kowska-Hoyer B, Jasik P, et al. Serum uromodulin concentrations
correlate with glomerular filtration rate in patients with chronic kidney
disease. Pol Arch Med Wewn 2016; 126:995–1004.
12. Steubl D, Block M, Herbst V, Nockher WA, Schlumberger W, Satanov-
skij R, et al. Plasma uromodulin correlates with kidney function and
identifies early stages in chronic kidney disease patients. Medicine
(Baltimore) 2016; 95:e3011.
13. Youhanna S, Weber J, Beaujean V, Glaudemans B, Sobek J, Devuyst O.
Determination of uromodulin in human urine: influence of storage and
processing. Nephrol Dial Transplant 2014; 29:136–145.
14. Onoe T, Yamada K, Mizushima I, Ito K, Kawakami T, Daimon S, et al.
Hints to the diagnosis of uromodulin kidney disease. Clin Kidney J
2016; 9:69–75.
15. El Achkar TM, Wu XR. Uromodulin in kidney injury: an instigator,
bystander, or protector? Am J Kidney Dis 2012; 59:452–461.
16. Rein P, Vonbank A, Saely CH, Beer S, Jankovic V, Boehnel C, et al.
Relation of albuminuria to angiographically determined coronary
arterial narrowing in patients with and without type 2 diabetes mellitus
and stable or suspected coronary artery disease. Am J Cardiol 2011;
107:1144–1148.
17. Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL
Jr, et al. The seventh report of the Joint National Committee on
Prevention, Detection, Evaluation, and Treatment of High Blood
Pressure: the JNC 7 report. JAMA 2003; 289:2560–2572.
18. The Renal Association. The UK eCKD guide. 2009; http://www.renal.
org/information-resources/the-uk-eckd-guide/#sthash.TXzTh2gI.dpbs.
19. National Kidney Foundation. K/DOQI clinical practice guidelines for
chronic kidney disease: evaluation, classification, and stratification. Am
J Kidney Dis 2002; 39:S1–S266.
20. Voight BF, Kang HM, Ding J, Palmer CD, Sidore C, Chines PS, et al. The
metabochip, a custom genotyping array for genetic studies of meta-
bolic, cardiovascular, and anthropometric traits. PLoS Genet 2012;
8:e1002793.
21. Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and
visualization of LD and haplotype maps. Bioinformatics 2005;
21:263–265.
22. Hlatky MA, Greenland P, Arnett DK, Ballantyne CM, Criqui MH, Elkind
MS, et al. Criteria for evaluation of novel markers of cardiovascular risk:
a scientific statement from the American Heart Association. Circulation
2009; 119:2408–2416.
23. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, Müller M.
pROC: an open-source package for R and Sþ to analyze and compare
ROC curves. BMC Bioinformatics 2011; 12:77.
24. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D,
et al. PLINK: a tool set for whole-genome association and population-
based linkage analyses. Am J Hum Genet 2007; 81:559–575.
25. Thornley C, Dawnay A, Cattell WR. Human Tamm–Horsfall glycopro-
tein: urinary and plasma levels in normal subjects and patients with
renal disease determined by a fully validated radioimmunoassay. Clin
Sci (Lond) 1985; 68:529–535.
26. Dawnay AB, Cattell WR. Serum Tamm–Horsfall glycoprotein levels in
health and in renal disease. Clin Nephrol 1981; 15:5–8.
27. Leiherer A, Muendlein A, Saely CH, Kinz E, Brandtner EM, Fraunberger
P, Drexel H. Serum uromodulin is associated with impaired glucose
metabolism. Medicine (Baltimore) 2017; 96:e5798.
28. Bichler KH, Ideler V, Harzmann R. Uromucoid excretion in
normal individuals and stone formers. Curr Probl Clin Biochem 1979;
309–324.
29. Lynn KL, Marshall RD. Excretion of Tamm–Horsfall glycoprotein in
renal disease. Clin Nephrol 1984; 22:253–257.
30. El Achkar TM, McCracken R, Rauchman M, Heitmeier MR, Al-Aly Z,
Dagher PC, Wu XR. Tamm–Horsfall protein-deficient thick
ascending limbs promote injury to neighboring S3 segments in an
MIP-2-dependent mechanism. Am J Physiol Renal Physiol 2011; 300:
F999–F1007.
31. El Achkar TM, Wu XR, Rauchman M, McCracken R, Kiefer S, Dagher PC.
Tamm–Horsfall protein protects the kidney from ischemic injury by
decreasing inflammation and altering TLR4 expression. Am J Physiol
Renal Physiol 2008; 295:F534–F544.
32. Jian L, Fa X, Zhou Z, Liu S. Functional analysis of UMOD gene and its
effect on inflammatory cytokines in serum of essential hypertension
patients. Int J Clin Exp Pathol 2015; 8:11356–11363.
www.jhypertension.com 117
 Leiherer et al.
33. Lhotta K. Uromodulin and chronic kidney disease. Kidney Blood Press
Res 2010; 33:393–398.
34. Delgado GE, Kleber ME, Scharnagl H, Krämer BK, März W, Scher-
berich JE. Serum uromodulin and mortality risk in patients
undergoing coronary angiography. J Am Soc Nephrol 2017;
28:2201 – 2210.
35. Zhou J, Chen Y, Liu Y, Shi S, Wang S, Li X, et al. Urinary uromodulin
excretion predicts progression of chronic kidney disease resulting from
IgA nephropathy. PLoS One 2013; 8:e71023.
36. Nkuipou-Kenfack E, Duranton F, Gayrard N, Argilés À,
Lundin U, Weinberger KM, et al. Assessment of metabolomic
and proteomic biomarkers in detection and prognosis of
progression of renal function in chronic kidney disease. PLoS One
2014; 9:e96955.
37. Steubl D, Block M, Herbst V, Schlumberger W, Nockher A, Angermann
S, et al. Serum uromodulin predicts graft failure in renal transplant
recipients. Biomarkers 2017; 22:171–177.
38. Leiherer A, Muendlein A, Saely CH, Ebner J, Brandtner EM, Fraun-
berger P, Drexel H. Serum uromodulin is a predictive biomarker for
cardiovascular events and overall mortality in coronary patients. Int J
Cardiol 2017; 231:6–12.
39. Lynn KL, Marshall RD. Excretion of Tamm–Horsfall glycoprotein in
renal disease. Clin Nephrol 1984; 22:253–257.
40. Ahluwalia TS, Lindholm E, Groop L, Melander O. Uromodulin gene
variant is associated with type 2 diabetic nephropathy. J Hypertens
2011; 29:1731–1734.
41. Prajczer S, Heidenreich U, Pfaller W, Kotanko P, Lhotta K, Jennings P.
Evidence for a role of uromodulin in chronic kidney disease progres-
sion. Nephrol Dial Transplant 2010; 25:1896–1903.
Reviewers’ Summary Evaluations
Reviewer 1
The authors evaluated the association between serum uro-
modulin and kidney function in 529 patients undergoing
coronary angiography. They found that serum uromodulin
concentration at baseline was significantly lower in patients
who developed CKD during the 4-year follow-up as com-
pared to those who did not; serum uromodulin concentra-
tion was inversely associated with the decline of eGFR and
significantly increased the performance of a prediction
model for CKD. Discussion fairly addresses the limitation
of the study and the relatively lack of knowledge about
uromudulin pathophysiology. Interestingly, the predictive
power of Uromodulin levels is mantained even after adjust-
met for baseline eGFR values.
118 www.jhypertension.com
Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
View publication stats
42. Rasch R, Torffvit O, Bachmann S, Jensen PK, Jacobsen NO. Tamm–
Horsfall glycoprotein in streptozotocin diabetic rats: a study of kidney
in situ hybridization, immunohistochemistry, and urinary excretion.
Diabetologia 1995; 38:525–535.
43. Thulesen J, Jorgensen PE, Torffvit O, Nexo E, Poulsen SS. Urinary
excretion of epidermal growth factor and Tamm–Horsfall protein in
three rat models with increased renal excretion of urine. Regul Pept
1997; 72:179–186.
44. Pfleiderer S, Zimmerhackl LB, Kinne R, Manz F, Schuler G, Brandis M.
Renal proximal and distal tubular function is attenuated in diabetes
mellitus type 1 as determined by the renal excretion of alpha 1-micro-
globulin and Tamm–Horsfall protein. Clin Investig 1993; 71:972–977.
45. Torffvit O, Agardh CD. Urinary excretion rate of NC1 and Tamm–
Horsfall protein in the microalbuminuric type I diabetic patient. J
Diabetes Complications 1994; 8:77–83.
46. Zimmerhackl LB, Pfleiderer S, Kinne R, Manz F, Schuler G, Brandis M.
Tamm–Horsfall-protein excretion as a marker of ascending limb
transport indicates early renal tubular damage in diabetes mellitus
type I. J Diabet Complications 1991; 5:112–114.
47. Bachmann S, Metzger R, Bunnemann B. Tamm–Horsfall protein-
mRNA synthesis is localized to the thick ascending limb of Henle’s
loop in rat kidney. Histochemistry 1990; 94:517–523.
48. Malagolini N, Cavallone D, Serafini-Cessi F. Intracellular transport, cell-
surface exposure and release of recombinant Tamm–Horsfall glyco-
protein. Kidney Int 1997; 52:1340–1350.
49. Padmanabhan S, Melander O, Johnson T, Di Blasio AM, Lee WK,
Gentilini D, et al. Genome-wide association study of blood pressure
extremes identifies variant near UMOD associated with hypertension.
PLoS Genet 2010; 6:e1001177.
Reviewer 2
In this study, the main aim was to examine the association
between serum uromodulin and renal function. The
authors identified that developing CKD was inversely as-
sociated with serum uromodulin concentration, indepen-
dent from other risk factors including patients’ blood
pressure, diabetes, CAD, as well as kidney markers ACR
and eGFR, and a stratification of the study population
according to the UMOD coding SNP rs13335818, in LD
with the promoter SNP rs12917707 previously identified
in GWAS. This highlights the potentially important role of
uromodulin in disease and emphasizes the critical require-
ment for a greater understanding of the biological role of
serum and urinary uromodulin in the general population
and patients.
Volume 36  Number 1  January 2018