Received: 7 February 2019 Revised: 2 April 2019
DOI: 10.1002/dmrr.3195
RESEARCH ARTICLE
Circulating TNF receptors 1 and 2 predict progression of
diabetic kidney disease: A meta‐analysis
Xiaoqi Ye | Ting Luo | Kanran Wang
Department of Endocrinology, The First
Affiliated Hospital of Chongqing Medical
University, Chongqing, China
Correspondence
Qifu Li and Jinbo Hu, Department of
Endocrinology, The First Affiliated Hospital of
Chongqing Medical University, No. 1 Youyi
Street, Yuzhong District, Chongqing, China,
400016.
Email: liqifu@yeah.net; hujinbo_568@163.com
Funding information
National key research & development plan,
major project of precision medicine research,
Grant/Award Numbers: 2017YFC0909600,
2017YFC0909602 and 2017YFC0909603;
National Natural Science Foundation of China,
Grant/Award Numbers: 81800731, 81700754
and 81670785
Co‐first authors: Xiaoqi Ye and Ting Luo.
Diabetes Metab Res Rev. 2019;e3195.
https://doi.org/10.1002/dmrr.3195
Accepted: 3 June 2019
| Yue Wang | Shumin Yang | Qifu Li | Jinbo Hu
Abstract
Background: We conducted a meta‐analysis to investigate the association of
circulating tumor necrosis factor‐1 (TNFR‐1) and TNFR‐2 with diabetic kidney dis-
ease (DKD) progression, which is the first‐ever quantitative analysis of these associ-
ations thus far. Whether TNFRs were better than albumin‐creatinine ratio (ACR) in
predicting DKD progression was also explored.
Methods: A systematic search of the PubMed, EMBASE, and Cochrane Library
databases up to 1 February 2018, was conducted. The main outcome was DKD pro-
gression, which was defined as eGFR decline, macroalbuminuria, or incidence of
DKD‐related events. Eligible studies were included for pooled analysis using either
fixed‐effects or random‐effects models to incorporate between‐study variation by
different measurement standards. Publication bias was evaluated using Egger's test.
Results: The meta‐analysis included 6526 participants from 11 cohorts with circu-
lating TNFR‐1 measurements and 5385 participants from 10 prospective studies with
circulating TNFR‐2 measurements. Compared with the lowest level category, diabetic
patients with the highest TNFR‐1 or TNFR‐2 level category exhibited a higher risk of
DKD progression (RR 2.51, 95% CI [1.92‐3.27] for TNFR‐1; 3.23 [1.99‐5.26] for
TNFR‐2). The risk of DKD progression was also increased with the per unit increment
of TNFR‐1 or TNFR‐2 (1.68 [1.43‐1.97] for TNFR‐1; 1.69 [1.31‐2.17] for TNFR‐2).
Although existing studies did not support a direct comparison between ACR and
TNFRs, it was undeniable that TNFRs could improve the predictive value in DKD
progression.
Conclusions: Circulating TNFR‐1 and TNFR‐2 are reliable predictors of DKD pro-
gression. Whether TNFRs are better than ACR at predicting DKD progression needs
to be further investigated.
K E Y W OR D S
DKD, meta‐analysis, prospective cohort, TNFR‐1, TNFR‐2
wileyonlinelibrary.com/journal/dmrr © 2019 John Wiley & Sons, Ltd. 1 of 8
2 of 8
1 | I N T RO D U CT I O N
Diabetic kidney disease (DKD) is one of the major microvascular compli-
1 HR and its 95% CI; and (d) included outcomes of macroalbuminuria,
cations of diabetes. DKD has also become the leading cause of chronic
kidney disease (CKD) and end‐stage renal disease (ESRD) around the
2,3
world. Albuminuria and estimated glomerular filtration rate (eGFR)
are currently the standard evaluation indicators for DKD. Nonetheless,
the so‐called “non‐albuminuric renal impairment” is not uncommon in
the clinic, which makes it less reliable for DKD progression.4
Tumor necrosis factor (TNF)‐α is an important mediator of renal
chronic inflammation, which is deemed as the key to DKD development.
The signal transduction of TNF‐α commences viaTNF receptor (TNFR)‐
1 and TNFR‐2, which are present in both membrane‐bound forms and
5 sensus with a third investigator.
soluble forms in serum. Some prospective studies suggested that a
higher level of circulating TNFR‐1 or TNFR‐2 was associated with an
increased risk of DKD progression.6-9 However, in a cohort study, nei-
ther circulating TNFR‐1 nor TNFR‐2 was significantly related to the risk
of all‐cause mortality in patients with diabetes.10 Circulating TNFR‐2
levels also failed to predict renal progression in another cohort of type
2 diabetes.11 It is necessary to further ascertain whether circulating
TNFR‐1/2 could be reliable predictors of DKD progression and were
better than albuminuria in predicting DKD progression.
In this systematic review and meta‐analysis, we aimed to estimate
the relationships between circulating TNFR‐1/2 and the risk of DKD
progression among prospective studies.
2 | METHODS
2.1 | Definition of DKD progression
DKD progression was defined as eGFR decline, macroalbuminuria
(albumin‐creatinine ratio [ACR] > 300 mg/g or albumin excretion rates
[AER] > 300 mg/24 h), or incidence of DKD‐related events. The eGFR
decline included eGFR<60 mL/min per 1.73 m2 or decline in eGFR of
≥25% from baseline. The DKD‐related events included all‐cause mor-
tality, ESRD (defined by the initiation of maintenance dialysis, receipt
of kidney transplant, or an eGFR of <15 mL/min per 1.73 m2), and
doubling of serum creatinine levels. In subgroup analysis, we further
defined eGFR decline or macroalbuminuria as early DKD progression.
And ESRD or doubling of serum creatinine levels was defined as
advanced DKD progression.
2.2 | Study selection
We conducted a systematic literature search by identifying all studies
that investigated the association between TNFR levels and DKD, pub-
lished before February 2018 in the PubMed, EMBASE, and Cochrane
Library databases. The electronic databases were explored using a
combination of the following terms: tumor necrosis factor receptor,
TNFR, diabetic patients, diabetes mellitus, DM, diabetic nephropathy,
albuminuria, eGFR, DKD, and diabetic kidney disease. Studies written
in all languages without any special restriction were included.
YE ET AL.
We included an article if it (a) was a prospective cohort study or a
nested case‐control study; (b) evaluated the association between the
elevated levels of TNFRs and DKD risk; (c) reported the OR, RR, or
decline of eGFR, a doubling increase in serum creatinine concentra-
tion, ESRD, or all‐cause mortality. In these studies, patients were
excluded if they did not have diabetes at baseline or had evidence
of nephropathy unrelated to diabetes. We also excluded duplicated
studies and reports, cross‐sectional studies, personal experience sum-
maries, literature reviews, studies on cell lines or animals, and data
displayed as figures or no relevant data reported. Two independent
reviewers evaluated articles eligible for further review after an initial
screen of abstracts or titles. Any disagreement was resolved by con-
2.3 | Data extraction and quality assessment
Two of the reviewers independently extracted the relevant data from
the databases according to a standardized form. The data for each
included article were as follows: first author's name, proportion of
women, publication year, journal name, sample size, DKD stage, bio-
marker measurements (assay), country in which the study was per-
formed, study design, mean age, duration of follow‐up, outcome,
reported adjusted odds ratios (ORs), relative risks (RRs), or hazard
ratios (HRs) of DKD comparing various levels of TNFRs and the corre-
sponding 95% CIs, and adjustment for the major confounding factors
in the design or data analysis. We extracted data by examining the full
text and any supplementary materials. If available, the risk estimate
that was most fully adjusted for the most variables was extracted.
The Newcastle‐Ottawa Scale (NOS) was used to assess the quality
of all eligible studies.12
2.4 | Statistical analysis
The multivariable‐adjusted HRs, RRs, or ORs reported in the eligible
studies were extracted and considered directly as RR in our analysis.
We used the most‐adjusted HR, RR, or OR and its 95% CI in all anal-
yses for a summary RR with its 95% CI. Pooled estimates of the RR
were used as measures of association between TNFR levels and the
risk of DKD progression. We stratified the analysis by the different
measurement standards. First, multivariable‐adjusted RRs and 95%
CIs were used to obtain the overall RR for the highest TNFR level cat-
egory compared with the lowest level category. Subsequently, RRs
and 95% CIs for each standard deviation (SD) or interquartile range
(IQR) increase in standardized TNFRs were used to obtain the overall
RR for the per unit increase.
We examined the heterogeneity in results across studies using
Cochran's Q and I2 statistics.13,14 We selected a random‐effect model
to calculate the overall RR value when I2 was more than 30%; otherwise,
a fixed‐effect model was used in the absence of heterogeneity.13,15 To
explore the source of heterogeneity and test the robustness of our
results, subgroup analyses were carried out by country, DKD stage
YE ET AL.
(early, late), duration of follow‐up (≥5 years, <5 years), and sample size
(≥500 people, <500 people). In addition, to assess the influence of indi-
vidual studies on the pooled result, we conducted a sensitivity analysis
by excluding each study one by one and recalculating the combined esti-
mates on the remaining studies. We used Egger's test (linear regression
method)16 to evaluate publication bias, and P < .05 for Egger's test was
considered to be representative of a significant statistical publication
bias. All statistical analyses were performed using Stata 14.1 (Stata‐
Corp, College Station, TX).
3 | RESULTS
We initially retrieved 1247 citations from the database. Of these,
1223 articles were excluded after reading the titles and abstracts
due to them not being relevant to DM, review and editorial articles,
and cell line or animal studies. After full‐text review of 24 articles, nine
studies were excluded because they were missing sufficient data,
cross‐sectional design, or did not identify the diabetic population.
Figure 1 shows the flow diagram of study selection for the analysis.
6,9,17,18
Four studies were excluded because they did not use appropri-
ate units for exposure data to provide pooled risk ratios. Finally, the
remaining 11 studies with 12 cohorts were included in our meta‐
analysis. The specific characteristics of the 11 studies are shown in
Table 1..7,8,10,11,19-25 Overall, the analysis included 6526 participants
from 11 cohorts with TNFR‐1 measurements and 5385 participants
from 10 prospective studies with TNFR‐2 measurements. For most
original studies, TNFR‐1 and TNFR‐2 levels were classified by tertiles
and quartiles. The fourth quartile or third tertile compared with the
lowest quartile or tertile was considered to be the highest TNFR level
category compared with the lowest category.
FIGURE 1 Flow chart of literature search
and study selection
3 of 8
3.1 | Meta‐analysis of TNFR‐1 and DKD progression
Of the 11 cohorts with TNFR‐1 measurements, six cohorts evaluated
risks of DKD progression by comparing the highest TNFR‐1 level cat-
egory with the lowest, and five cohorts provided the RRs of DKD pro-
gression for the per unit increment of TNFR‐1 level. For cohorts with
the level category comparison, a higher TNFR‐1 level was associated
with an increased risk of DKD progression (RR 2.51, 95% CI 1.92‐
3.27) (Figure 2). The between‐study heterogeneity was not significant
among studies (Pheterogeneity = 0.359; I2 = 9%). No publication bias was
found by the Egger's test (P = .746) (Figure S4). A sensitivity analysis
was also performed to confirm the robustness of our findings. We
recalculated the pooled risk estimates for the remainder of the studies
by omitting one study at a time, which resulted in little change in the
observed risk estimates from 2.32 (95% CI 1.72‐3.14) to 2.93 (95% CI
2.17‐3.95) (Figure S3).
In cohorts providing the RRs of DKD progression for the per unit
increment of TNFR‐1 level, the pooled risk estimate was 1.68 (95%
CI 1.43‐1.97) (Figure 2). Between‐study heterogeneity was found
among studies (Pheterogeneity = 0.056, I2 = 56.5%). There was no indica-
tion of a publication bias from the result of Egger's test (P = .907) (Fig-
ure S4). In sensitivity analyses, the observed risk estimates ranged
from 1.60(95% CI 1.41‐1.82) to 1.76(95% CI 1.31‐2.07) (Figure S3).
Also, when we excluded the study for IQR increase, the result did
not change (RR 1.69, 95% CI 1.40‐2.05), even with a higher
heterogeneity.
To test the robustness of our results, we conducted subgroup anal-
yses. The results showed that TNFR‐1 was associated with both early
and advanced DKD progression (Figure 3, Figure S1), which was not
markedly influenced by country, DKD stage, duration of follow‐up,
or sample size (Table S1).
Colour online, B&W in print
4 of 8 YE ET AL.

TABLE 1 Study characteristics and RRs of DKD associated with TNFR‐1 and/or TNFR‐2

Follow‐
DM Sample Up DKD Unit of Adjusted NOS
Author Year Design Country Type Size duration Assay Stage Outcome Biomarker RR(95% CI) Score

Coca (ACCORD) 2017 NC America T2DM 380 5 ELISA Early eGFR decline H/L TNFR‐1 2.99 8
et al (1.23‐7.25)
TNFR‐2
8.37(3.00‐
23.41)
Coca 2017 C America T2DM 1256 2.2 ELISA Late DKD‐related H/L TNFR‐1 3.45 8
(NEPHRON‐D) events (1.9‐6.27)
et al TNFR‐2 3.80
(1.99‐7.26)
Lin et al 2010 C America T2DM 516 11 ELISA Early eGFR decline H/L TNFR‐2 5.81 7
(2.9‐11.65)
Fernández et al 2017 C Spain T2DM 101 2.7 ELISA Late DKD‐related H/L TNFR‐1 2.6 6
events (1.11‐6.34)
TNFR‐2 1.8
(0.8‐3.8)
Saulnier et al 2014 C France T2DM 522 4 ELISA Late DKD‐related H/L TNFR‐1 2.98 7
events (1.7‐5.23)
Gohda et al 2012 C America T1DM 628 12 ELISA Early eGFR decline H/L TNFR‐1 2.53 7
(1.36‐4.67)
TNFR‐2 3.04
(1.67‐5.52)
Gohda et al 2017 C Japan / 160 4.4 ELISA Late DKD‐related H/L TNFR‐1 1.4 6
events (0.78‐2.49)
TNFR‐2 1.51
(0.82 2.78)
Virella et al 2013 C America T1DM 1237 6.5 Signature Early Macroalbuminuria PUI TNFR‐1 1.33 8
plus (1.01‐1.76)
protein TNFR‐2 1.43
(1.07‐1.91)
Saulnier et al 2017 C France T2DM 1135 4.3 ELISA Early eGFR decline PUI TNFR‐1 1.78 8
(1.57‐2.02)
Carlsson et al 2016 C Sweden T2DM 607 7.4 ELISA Late DKD‐related PUI TNFR‐1 1.49 7
events (1.19‐1.86)
TNFR‐2 1.51
(1.18‐1.93)
Pavkov et al 2014 C America T2DM 193 9.5 ELISA Early DKD‐related PUI TNFR‐1 1.6 8
events (1.1‐2.2)
TNFR‐2 1.7
(1.2‐2.3)
Looker et al 2015 C Scotland T2DM 307 3.5 ELISA Late eGFR decline PUI TNFR‐1 2.41 7
(1.76‐3.37)
TNFR‐2 2.55
(1.84‐3.63)

* NC: nested case‐control. C: cohort. T1DM: type 1 diabetes mellitus. T2DM: type 2 diabetes mellitus. eGFR decline: defined as eGFR<60 mL/min per
1.73 m2, or decline in eGFR of ≥25% from baseline. Macroalbuminuria: defined as urine albumin excretion rate ≥ 300 mg/24 h. DKD‐related events:
included all‐cause mortality, or end stage renal disease (ESRD), or doubling of serum creatinine levels. H/L: defined as the highest TNFR level category com-
pared with the lowest level category. PUI: defined as per unit increase in TNFR levels. DKD stage: early defined eGFR>60 mL/min per 1.73 m2 or free of
macroalbuminuria at baseline; late defined eGFR<60 mL/min per 1.73 m2 or macroalbuminuria at baseline. ELISA: enzyme‐linked immunosorbent assay.

3.2 | Meta‐analysis of TNFR‐2 and DKD progression
Of the 10 cohorts with TNFR‐2 measurements, six cohorts evaluated
risks of DKD progression by comparing the highest TNFR‐2 level cat-
egory with the lowest, and four cohorts provided the RRs of DKD
progression for the per unit increment of TNFR‐2 level. For cohorts
with the level category comparison, a higher TNFR‐2 level was associ-
ated with a greater risk of DKD progression (RR 3.23, 95% CI 1.99‐
5.26) (Figure 2). Between‐study heterogeneity was found among stud-
ies (Pheterogeneity = 0.014; I2 = 65.1%). Egger's test showed that no
YE ET AL.
FIGURE 2 Forest plot of TNFR levels and the risk of DKD progression for different measurement standards. The rectangle represents the point
estimate (horizontal line indicates the 95% CI), with its size being proportional to the weight of the study in the meta‐analysis. The diamond
represents the pooled estimate (with its size representing the 95% CI) ONLY
publication bias existed in this meta‐analysis (P = .329) (Figure S4). We
performed the sensitivity analysis by excluding individual studies and
found that the results remained consistent, with the observed risk
estimates ranging from 2.83 (95% CI 1.76‐4.56) to 3.80 (95% CI
2.42‐5.98) (Figure S3).
In cohorts providing the RRs of DKD progression for the per unit
increment of TNFR‐2 level, the overall RR of DKD was 1.69 (95% CI
1.31‐2.17) for the per unit increment in TNFR‐2 levels (Figure 2). Sub-
2
stantial between‐study heterogeneity (Pheterogeneity = 0.022; I = 69%)
was found in the analysis. Meanwhile, no publication bias was found
by Egger's test (P = .128) (Figure S4). Exclusion of any single study
from the analysis did not alter the overall finding in a sensitivity test,
which resulted in a range of the observed risk estimates from 1.48
(95% CI 1.27‐1.72) to 1.84 (1.36‐2.50) (Figure S3).
In the subgroup analyses, no association between TNFR‐2 and
the risk of early DKD progression was found for the per unit incre-
ment of TNFR‐2 level, while it might not be convincing for its high
between‐study heterogeneity (Figure S1). There was no marked
5 of 8
Colour online, B&W in print
influence in duration of follow‐up, DKD stage, and sample size
(Table S2).
3.3 | TNFR‐1/2 vs ACR on the risk of DKD
progression
To further determine whether TNFRs performed better than albumin-
uria in predicting DKD progression, we also conducted a subgroup anal-
ysis for the results of TNFRs and albuminuria in the same studies.
Because the units used for the exposure were different in each study,
it was difficult to calculate a pooled risk ratio, but forest plots were cre-
ated (Figure S2). We also searched for studies that included the c‐
statistic, integrated discriminant index (IDI), or net reclassification index
(NRI) to explore the added value of TNFRs as predictors. We found
three studies that assessed the improvement in risk discrimination for
TNFRs compared with the model with ACR (Table S3). All of the studies
demonstrated incremental predictive value of TNFRs to ACR.
 6 of 8 YE ET AL.
Colour online, B&W in print

FIGURE 3 Forest plot of TNFR levels and the risk of advanced DKD progression for different measurement standards. The advanced DKD was
defined as end stage renal disease (ESRD) and doubling of serum creatinine levels. The rectangle represents the point estimate (horizontal line
indicates the 95% CI), with its size being proportional to the weight of the study in the meta‐analysis. The diamond represents the pooled estimate
(with its size representing the 95% CI)

4 | DISCUSSION
This meta‐analysis included 6526 participants from 11 cohorts with
TNFR‐1 measurements and 5385 participants from 10 prospective
studies with TNFR‐2 measurements, which showed that elevated con-
centrations of TNFR‐1 or TNFR‐2 were independently associated with
a higher risk of DKD progression. These findings further supported the
hypothesis that TNFR‐1 and TNFR‐2 were reliable predictors of DKD
progression.
The association between TNFR‐1 and the risk of DKD progression
was reported by many previous studies. In a French cohort of 1135
patients with type 2 diabetes who were monitored for 11.8 years,
TNFR‐1 was shown to independently predict a decline in eGFR-
40% compared with the baseline value.23 In another cohort that
contained 522 patients with DKD (eGFR<60 mL/min/1.73 m2
and/or UACR>30 mg/g), TNFR‐1 proved to be a strong prognostic
21
factor for all‐cause mortality. However, Gohda et al reported that
in a cohort that included 319 patients receiving maintenance hemodi-
alysis, circulating TNFR‐1 was not associated with the risk of all‐cause
mortality in patients with diabetes.10 In Gohda's study, the DKD pro-
gression was a subgroup analysis of only 50% of subjects who were
diabetic patients, so the results should be cautiously interpreted. In
our study, both meta‐analyses of level category comparison and RRs
for the per unit increment showed a positive correlation between
TNFR‐1 and DKD progression, which strongly indicated that TNFR‐1
could be a reliable predictor of DKD progression.
A tight relationship between TNFR‐2 and DKD progression has
also been reported. In a cohort of 516 women with type 2 diabetes,
comparing the highest with the lowest quartile, TNFR‐2 was indepen-
dently associated with an eGFR decline of ≥25% (multivariate OR
5.81; 95% CI 2.90‐11.65).20 Another cohort of 349 diabetic patients
with proteinuria was followed‐up for 5 to 18 years, and the results
showed that TNFR‐2 was the strongest determinant of renal decline
and ESRD risk (C‐index 0.79).26 However, in a cohort including 101
patients with DKD, the circulating TNFR‐2 level failed to predict renal
progression (defined as >50% increase in serum creatinine concentra-
tion, end‐stage renal disease [ESRD], or death).11 As this study was a
randomized controlled trial primarily aimed at comparing the effect
of optimal doses of renin‐angiotensin system blockers, the negative
relationship between TNFR‐2 and DKD progression might not be con-
vincing. In our meta‐analysis, a positive correlation between TNFR‐2
and DKD progression was confirmed by both the method of level cat-
egory comparison and RRs for the per unit increment, which proved
that TNFR‐2 was effective in the prediction of DKD progression.
YE ET AL. 7 of 8

It should be mentioned that main heterogeneity was detected in the excluded because they did not use common unit form, and the original
meta‐analysis for the per unit increment of TNFRs level, but it data were not successfully obtained.
decreased (I2 from 69% to 0%) with the removal of one study reported In summary, this meta‐analysis indicate that either TNFR‐1 or
25
by Looker et al. In this study, patients with baseline eGFR 30 to TNFR‐2 can be an independent biomarker of DKD progression. To
60 mL/min per 1.73 m2 were included, so the heterogeneity may be determine whether TNFRs are better than albuminuria, further well‐
due to a relative narrow range of renal function. Not only that, the het- designed cohort and well‐planed intervention studies should be car-
erogeneity may be attributed to sample sizes, populations, measure- ried out to validate our findings.
ment errors, and so on. Moreover, the sensitivity analysis indicated
the reliability of the results and no publication bias in this meta‐analysis.
FINANCIAL SUPPORT INFORMATION
Albuminuria is now widely used as a predictor of DKD progression
in the clinic. A meta‐analysis selected cohort studies of diabetic
National key research &development plan, major project of precision
patients that provided adjusted RR of albuminuria for risks of all‐cause
medicine research (2017YFC0909600, sub‐project:
mortality and renal events (defined as renal replacement therapy, renal
2017YFC0909602, 2017YFC0909603); and the National Natural Sci-
transplantation, or loss of renal function), and either microalbuminuria
ence Foundation of China (81670785) to Qifu Li; the National Natural
or macroalbuminuria was a significant risk factor compared with
Science Foundation of China (81800731) to Jinbo Hu; the National
normoalbuminuria.27 However, the association of microalbuminuria
Natural Science Foundation of China (81700754) to Ting Luo.
with progressive renal disease has been highly challenged, as the pro-
gression of DKD does not always accompany an increase of albumin-
ACKNOWLEDGEMENTS
uria, and patients with normal albuminuria could also develop
ESRD.28,29 In a meta‐analysis including 1 024 977 participants The authors thank the First Affiliated Hospital of Chongqing Medical
(128 505 with diabetes), microalbuminuria did not predict end‐stage University.
renal disease in chronic kidney disease populations with diabetes,
while macroalbuminuria did.30 In a cohort including 628 T1DM with CONFLIC T OF INT E RE ST
high‐normal AER (15‐29 mg/min) or microalbuminuria (30‐300 mg/
The authors have no conflicts of interest to report.
min), AER failed to predict renal progression (defined as
eGFR<60 mL/min per 1.73 m2), while TNFRs were strong predictors.8
AUTHOR CONT R IBUT IONS
However, TNFR1 and TNFR2 as potential biomarkers for the pre-
diction of DKD progression, the limitations should not be ignored, Xiaoqi Ye and Ting Luo are co‐first authors and contribute equally.
including the lack of large‐sample study, long‐term observation, and Xiaoqi Ye, Ting Luo, and Qifu Li designed the study, oversaw the data
direct head‐to‐head comparison of TNFRs with albuminuria in a pro- collection, and wrote the manuscript. Jinbo Hu and Kanran Wang con-
spective study. Hence, further comprehensive studies need to be tributed to the data analysis and the writing of the manuscript. Jinbo
designed and implemented in the future. Hu and Qifu Li contributed to critical revisions of the article for impor-
In DKD, other potential predictive biomarkers have also been tant intellectual content.
explored, including fibroblast growth factor 21, fibroblast growth fac-
tor 23 and pigment epithelium‐derived factor, kidney injury marker‐1, DISC LOSURE S UMMARY
transforming growth factor‐ β, connective tissue growth factor, and
neutrophil gelatinase‐associated lipocalin.18,25,31-34 Nevertheless, no All of the authors have nothing to disclose.
comprehensive studies have compared whether they performed bet-
ter than existing indicators (such as albuminuria and TGFRs) in ET HIC S ST AT EME NT
predicting DKD progression.
This study is a meta‐analysis. We followed ethical principles.
The main strength of our study was the longitudinal design. Our
study included a systematic search of all available published prospec-
ORCID
tive cohorts, which is the first‐ever quantitative synthesis of these
associations thus far. Furthermore, we used different measurement Jinbo Hu https://orcid.org/0000-0003-2925-0583
standards to evaluate the relationship between TNFR1/2 and the risk
of DKD, with more accuracy and rationality. There are several limita- RE FE RE NC ES
tions that should be taken into account. On one hand, the adjustments
1. Tuttle KR, Bakris GL, Bilous RW, et al. Diabetic kidney disease: a report
for potential confounders differed in the original studies. Although we from an ADA Consensus Conference. Diabetes Care.
used adjusted estimates from multivariate models in each study, the 2014;37(10):2864‐2883.
influence of possible confounders on the meta‐analysis cannot be
2. Wu MS, Wu IW, Shih CP, Hsu KH. Establishing a platform for battling
entirely excluded. On the other hand, different methods were adopted
endstage renal disease and continuing quality improvement in dialysis
to measure TNFRs (eg, different ELISA kits) among included studies, therapy in Taiwan Taiwan Renal Registry Data System (TWRDS). Acta
which may also influence the results. Moreover, some studies were Nephrologica. 2011;25:148‐153.
8 of 8
3. ERA‐EDTA Registry. ERA‐EDTA Registry Annual Report 2013. Aca-
demic Medical Center, Department of Medical Informatics,
Amsterdam, The Netherlands, 2015
4. Lin C, Chang Y, Chuang L. Early detection of diabetic kidney disease:
present limitations and future perspectives. World J Diabetes.
2016;7(14):290‐301.
5. Cabalhierro L, Lazo PS. Signal transduction by tumor necrosis factor
receptors. Cell Signal. 2012;24(6):1297‐1305.
6. Niewczas MA, Gohda T, Skupien J, et al. Circulating TNF receptors 1
and 2 predict ESRD in type 2 diabetes. J Am Soc Nephrol.
2012;23(3):507‐515.
7. Pavkov ME, Nelson RG, Knowler WC, Cheng Y, Krolewski AS,
Niewczas MA. Elevation of circulating TNF receptors 1 and 2 increases
the risk of end‐stage renal disease in American Indians with type 2 dia-
betes. Kidney Int. 2014;87(4):812‐819.
8. Gohda T, Niewczas MA, Ficociello LH, et al. Circulating TNF receptors 1
and 2 predict stage 3 CKD in type 1 diabetes. J Am Soc Nephrol.
2012;23(3):516‐524.
9. Krolewski AS, Niewczas MA, Skupien J, et al. Early progressive renal
decline precedes the onset of microalbuminuria and its progression to
macroalbuminuria. Diabetes Care. 2014;37(1):226‐234.
10. Gohda T, Maruyama S, Kamei N, et al. Circulating TNF receptors 1 and
2 predict mortality in patients with end‐stage renal disease undergoing
Dialysis. Sci Rep. 2017;7(1):43520.
11. Fernándezjuárez G, Perez JV, Fernández JL, et al. High levels of circu-
lating TNFR‐1 increase the risk of all‐cause mortality and progression
of renal disease in type 2 diabetic nephropathy. Nephrology (Carlton).
2017;22(5):354‐360.
12. Wells G, Shea B, O'Connell D, et al. The Newcastle‐Ottawa Scale (NOS)
for assessing the quality of nonrandomised studies in meta‐analyses.
Ottawa: Ottawa Hospital Research Institute; 2011.
13. DerSimonian R, Laird N. Meta‐analysis in clinical trials. Control Clin Tri-
als. 1986;7(3):177‐188.
14. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta‐
analysis. Stat Med. 2002;21(11):1539‐1558.
15. Su PF, Chen X, Chen H, Shyr Y. Statistical aspects of omics data anal-
ysis using the random compound covariate. BMC Syst Biol.
2012;6(Suppl 3):S11. https://doi.org/10.1186/1752-0509-6-S3-S11
16. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta‐
analysis detected by a simple, graphical test. BMJ.
1997;315(7109):629‐634.
17. Lee JE, Gohda T, Walker WH, et al. Risk of ESRD and all cause mortal-
ity in type 2 diabetes according to circulating levels of FGF‐23 and
TNFR‐1. Plos One. 2013;8(3): e58007. https://doi.org/10.1371/jour-
nal.pone.0058007
18. Barr ELM, Barzi F, Hughes JT, et al. High baseline levels of tumor necro-
sis factor receptor 1 are associated with progression of kidney disease in
Indigenous Australians with diabetes: the eGFR follow‐up study. Diabe-
tes Care. 2018;41(4):739, dc171919‐747.
19. Coca SG, Nadkarni GN, Huang Y, et al. Plasma biomarkers and kidney
function decline in early and established diabetic kidney disease. J Am
Soc Nephrol. 2017;28(9):2786‐2793.
20. Lin J, Hu FB, Mantzoros C, Curhan GC. Lipid and inflammatory bio-
markers and kidney function decline in type 2 diabetes. Diabetologia.
2010;53(2):263‐267.
21. Saulnier PJ, Gand E, Ragot S, et al. Association of serum concentration
of TNFR‐1 with all‐cause mortality in patients with type 2 diabetes and
chronic kidney disease: follow‐up of the SURDIAGENE cohort. Diabe-
tes Care. 2014;37(5):1425‐1431.
YE ET AL.
22. Lopesvirella MF, Baker NL, Hunt KJ, et al. Baseline markers of inflam-
mation are associated with progression to macroalbuminuria in type 1
diabetic subjects. Diabetes Care. 2013;36(8):2317‐2323.
23. Saulnier PJ, Gand E, Velho G, et al. Association of circulating bio-
markers (adrenomedullin, TNFR‐1, and NT‐proBNP) with renal
function decline in patients with type 2 diabetes: a French prospective
cohort. Diabetes Care. 2017;40(3):367‐374.
24. Carlsson AC, Östgren CJ, Nystrom FH, et al. Association of soluble
tumor necrosis factor receptors 1 and 2 with nephropathy, cardiovas-
cular events, and total mortality in type 2 diabetes. Cardiovasc
Diabetol. 2016;15(1):1‐8.
25. Looker HC, Colombo M, Hess S. Biomarkers of rapid chronic kidney
disease progression in type 2 diabetes. Kidney Int. 2015;88(4):888‐896.
26. Skupien J, Warram JH, Niewczas MA, et al. Synergism between circu-
lating tumor necrosis factor receptor 2 and HbA1c in determining renal
decline during 5–18 years of follow‐up in patients with type 1 diabetes
and proteinuria. Diabetes Care. 2014;37(9):2601‐2608.
27. Toyama T, Furuichi K, Ninomiya T, et al. The impacts of albuminuria and
low eGFR on the risk of cardiovascular death, all‐cause mortality, and
renal events in diabetic patients: meta‐analysis. Plos One. 2013;8(8):
e71810.
28. Caramori ML, Fioretto P, Mauer M. Low glomerular filtration rate in
normoalbuminuric type 1 diabetic patients: an indicator of more
advanced glomerular lesions. Diabetes. 2003;52(4):1036‐1040.
29. Yokoyama H, Sone H, Oishi M, Kawai K, Fukumoto Y, Kobayashi M.
Japan diabetes clinical data management study group. Prevalence of
albuminuria and renal insufficiency and associated clinical factors in
type 2 diabetes: the Japan Diabetes Clinical Data Management study
(JDDM15). Nephrol Dial Transplant. 2009;24:1212‐1219.
30. Mahmoodi BK, Matsushita K, Woodward M, et al. Associations of kid-
ney disease measures with mortality and end‐stage renal disease in
individuals with and without hypertension: a meta‐analysis. Lancet.
2012;380(9854):1662‐1673.
31. Antonellis PJ, Kharitonenkov A, Adams AC. Physiology and Endocrinol-
ogy Symposium: FGF21: insights into mechanism of action from
preclinical studies. J Anim Sci. 2014;92(2):407‐413.
32. Jian W‐X, Peng W‐H, Jin J, et al. Association between serum fibroblast
growth factor 21 and diabetic nephropathy. Metabolism.
2012;61(6):853‐859.
33. Hui E, Yeung C‐Y, Lee PCH, et al. Elevated circulating pigment
epithelium‐derived factor predicts the progression of diabetic
nephropathy in patients with type 2 diabetes. J Clin Endocrinol Metab.
2014;99(11):E2169‐E2177.
34. Colhoun HM, Marcovecchio ML. Biomarkers of diabetic kidney dis-
ease. Diabetologia. 2018;61(5):996‐1011.
SUPPORTING INFORMATION
Additional supporting information may be found online in the
Supporting Information section at the end of the article.
How to cite this article: Ye X, Luo T, Wang K, et al. Circulating
TNF receptors 1 and 2 predict progression of diabetic kidney
disease: A meta‐analysis. Diabetes Metab Res Rev. 2019;
e3195. https://doi.org/10.1002/dmrr.3195