Abstract
Background. Inflammation is a potential factor in the development and progression of diabetic nephropathy. The aim of this study was to analyse the relationship between the pro-inflammatory cytokine tumour necrosis factor-α (TNFα) and clinical markers of glomerular and tubulointerstitial damage [urinary albumin excretion (UAE) and urinary N-acetyl-β-glucosaminidase (UNAG), respectively] in a large group of type 2 diabetic patients.
Methods. A total of 160 diabetic patients and 32 healthy controls were included in the study. High-sensitive C-reactive protein (hs-CRP) as well as serum and urinary levels of TNFα were measured. UAE and UNAG were determined by 24-h urine collection.
Results. Serum hs-CRP and TNFα were significantly higher in diabetic than in control subjects, as well as UAE and UNAG. Diabetic patients had increased urinary TNFα compared to non-diabetics [14.5 (2–29) vs 4 (0.8–12), P < 0.001]. Serum hs-CRP and TNFα in diabetics with increased UAE were elevated compared to diabetics having normoalbuminuria. Urinary TNFα was also higher in diabetic subjects with micro- or macroalbuminuria than in patients with normal UAE [10.5 (4–20) and 18 (9–29) vs 7 (2–18) pg/mg, P < 0.0001, respectively]. Multiple regression analysis showed that urinary TNFα ( P < 0.0001), hs-CRP ( P < 0.0001), serum TNFα ( P < 0.01) and HbA1c ( P < 0.05) were independent of and significantly associated with UAE, whereas duration of diabetes ( P < 0.001), urinary TNFα ( P < 0.01), HbA1c ( P = 0.01), hs-CRP ( P < 0.05) and serum creatinine ( P < 0.05) were associated with UNAG.
Conclusions. In patients with type 2 diabetes, urinary TNFα excretion is elevated and correlates with severity of renal disease in terms of both glomerular and tubulointerstitial damage, suggesting a significant role for TNFα in the pathogenesis and progression of renal injury in diabetes mellitus.
Introduction
Urinary albumin excretion (UAE) is the classical marker of diabetic nephropathy (DN). However, tubulointerstitial involvement is also a major feature of renal lesion in diabetes, and more important, tubulointerstitial injury is a critical factor in the development and progression of renal dysfunction [ 1 ]. N-acetyl-β-glucosaminidase (NAG) is a lysosomal enzyme which originates from the proximal tubular cells. Increased urinary NAG excretion (UNAG) occurs in cases of significant tubular lesion in different diseases, including diabetes mellitus. Therefore, whereas UAE is a marker for glomerular injury, UNAG is widely used as a marker for tubular renal function [ 2 ].
Nowadays, there is growing evidence that activated innate immunity and inflammation are relevant factors in the pathogenesis of diabetes, with convincing data that type 2 diabetes includes an inflammatory component [ 3 ]. Regarding diabetic complications, experimental and clinical studies have demonstrated that DN exhibits signs of inflammation, with pro-inflammatory cytokines being suggested as important factors in the development of renal injury [ 4–7 ].
Tumour necrosis factor-α (TNFα) is a pleiotropic cytokine that plays an essential role in mediating inflammatory processes. Furthermore, TNFα is cytotoxic to glomerular, mesangial and epithelial cells, and may induce direct renal damage [ 8 , 9 ]. Studies in animal models of DN have shown that renal expression of TNFα is increased compared to kidneys of non-diabetic animals [ 10 ]. Preliminary clinical reports have found a significant relationship between the urinary excretion of this cytokine and UAE in type 2 diabetic patients [ 5 ]. However, the relationship between TNFα and tubulointerstitial damage has not been previously analysed.
The aim of the present study was to analyse the relationship between TNFα, one of the most important pro-inflammatory cytokines, and clinical markers of glomerular (UAE) and tubulointerstitial damage (UNAG) in a large group of type 2 diabetic patients.
Materials and methods
A total of 160 diabetic patients (83 men, 77 women; aged 62 ± 9 years) from our out-patient diabetic clinic were included in this study, as well as 32 healthy control subjects (15 men, 17 women; mean age 60 ± 8 years). The protocol was in accordance with the Declaration of Helsinki and the Convention on Human Rights and Biomedicine from the Council of Europe. Informed consent was obtained in all cases.
Patients were categorized as having normoalbuminuria when UAE was persistently <30 mg/day ( n = 25). Microalbuminuria was defined as an UAE between 30 and 300 mg/day ( n = 60). Finally, patients were categorized as having macroalbuminuria if they had persistent UAE >300 mg/day ( n = 75). Subjects with a current acute illness (including infectious diseases), severe proteinuria (UAE >3 g/day), blood pressure (BP) >160/100 mmHg, renal insufficiency (defined as a serum creatinine (Cr) >1.3 mg/dl), present cigarette smoking or medical history of clinical cardiovascular events in the last 6 months were excluded in order to avoid potential confounding factors. Before the definitive inclusion, the possible existence of active immunological diseases, malignancy and infections were investigated. White blood cell count was lower than 10 000/mm 3 in all cases. Tumoural markers including carcino-embryonic antigen, α-fetoprotein, cancer antigen 125 and prostate-specific antigen were negative. Serological tests for antinuclear antibodies, antineuthrophil cytoplasmic antibodies, cryoglobulins, rheumatoid factor, immunoglobulins and complement factors C3 and C4 were negative or within the normal range. Urine cultures and serology to hepatitis B and C and human immunodeficiency virus were also negative.
All diabetic patients were insulin dependent, had signs of diabetic retinopathy and were hypertensive under treatment with the maximum recommended doses of angiotensin-converting enzyme inhibitors (ACEIs) or angiotensin II type 1 receptor antagonists (ARA) for more than 6 months. BP was measured by a mercury sphygmomanometer with the patient in the sitting position after 5 min of rest. Three readings separated by 2 min were taken and the average was used for calculation. The first appearance of sound (phase 1) was used to define systolic BP. The disappearance of sound (phase 5) was used to define diastolic BP. Body mass index (BMI) was calculated as (weight) kg/(height) m 2 .
Laboratory analysis and cytokine measurements
Blood samples were drawn before breakfast in the morning (between 8 and 11 a.m.), after an 8–12-h overnight fast. Samples were collected in sterile tubes, centrifuged at 3000 g . for 10 min at 4°C, and then stored at −80°C until assayed. The plasma glucose level was measured by an automated enzymatic method. The glycated haemoglobin (HbA1c) concentration was measured by high performance liquid chromatography. The concentration of high-sensitive C-reactive protein (hs-CRP) was measured by ultrasensitive competitive immunoassay (Calbiochem, La Jolla, CA, USA) (interassay coefficient of variation of 8.4%). The serum TNFα concentration was determined by a high sensitivity quantitative sandwich enzyme immunoassay (R&D systems, Minneapolis, MN, USA) (lower limit of detection, 0.4 pg/ml; intra- and interassay coefficients of variation of the assay, 5.6 and 8.1%, respectively).
UAE was determined by 24-h urine collections. After collection, the samples were centrifuged at 3000 r.p.m. for 10 min and the supernatant was stored at −20°C. Urinary albumin was quantified by immunoturbidimetry (coefficient of variation 5.5%). Enzyme-linked immunosorbent assay (ELISA) was used for the detection of urinary TNFα. Urinary NAG was measured using a colorimetric assay with 3-cresolsulfonphthaleyn- n -acetyl-β- d -glucosaminide as a substrate (NAG colour test, Boehringer Mannheim), which is hydrolysed by NAG with the release of 3-cresolfulfonphthaleyn sodium salt, and measured photometrically at 580 nm. Urinary levels of TNFα and NAG were related to the concomitant urinary Cr concentration content in order to compensate for variations in urinary concentration, and expressed as NAG/Cr (U/g) and TNFα/Cr (pg/mg).
Statistical analysis
Results are presented as mean ± SD, except for UAE, UNAG, hs-CRP and TNFα, which are presented as median and range. Before analysis, these variables were logarithmically transformed. Differences in parameters between diabetic patients and control subjects were analysed by the Student's t -test and the Mann–Whitney U-test when appropriate. Comparison of patients stratified by albuminuria status was performed by one-way analysis of variance and Kruskal–Wallis analysis of variance. Post hoc comparisons were made using the Bonferroni procedure. Correlation between variables was calculated using the Pearson's and the Spearman correlation tests. Partial correlation analysis was performed to determine the extent to which a relationship was altered after adjusting for the other variables. Forward stepwise multiple regression analysis was performed to determine the independent association between relevant patient characteristics and clinical parameters as potential predictor variables (age, sex, duration of diabetes, BMI, systolic and diastolic BP, HbA1c, serum Cr, hs-CRP and serum and urinary TNFα) and UAE and UNAG as the dependent variables. Colinearity was excluded by the analysis of the eigenvalue. P -values <0.05 were considered statistically significant.
Results
General characteristics and clinical parameters of healthy non-diabetic controls and diabetic patients stratified by albuminuria status are depicted in Table 1 . Serum concentrations of the inflammatory parameters hs-CRP and TNFα as well as urinary TNFα were significantly higher in diabetic patients than in control subjects: 5.2 (0.8–9.3) mg/l, 5.9 (0.7–18) pg/ml and 14.5 (2–29) pg/mg vs 1.9 (0.2–3.6) mg/l, 2.3 (0.7–5) pg/ml and 4 (0.8–12) pg/mg, P < 0.001, respectively. Concerning urinary measurements, UAE and UNAG were also higher in diabetics than in healthy controls: 210 (6–2630) mg/day and 6.3 (1.7–88) U/g vs 13 (6–26) mg/day and 4.3 (0.8–11) U/g, P < 0.001 and P < 0.01, respectively.
General and clinical parameters in healthy controls and type 2 diabetic patients stratified by albuminuria status
| Controls ( n = 32) | Normoalbuminuria ( n = 25) | Microalbuminuria ( n = 60) | Macroalbuminuria ( n = 75) | |
|---|---|---|---|---|
| Age (years) | 60 ± 8 | 60 ± 6 | 59 ± 8 | 64 ± 9 * |
| Men/women | 17/15 | 13/12 | 29/26 | 36/33 |
| Duration of diabetes (years) | – | 8.3 ± 1.7 | 12.2 ± 3.4 ** | 11.9 ± 3.8 ** |
| BMI (kg/m 2 ) | 25.7 ± 2.3 | 25.3 ± 3.7 | 26.1 ± 2.4 | 25.6 ± 2.54 |
| Systolic BP (mmHg) | 131 ± 4 | 135 ± 3 † | 138 ± 8 † | 141 ± 9 †,†† |
| Diastolic BP (mmHg) | 76 ± 6 | 80 ± 6 † | 85 ± 8 † | 84 ± 9 † |
| Glucose (mg/dl) | 89 ± 7 | 161 ± 13 ‡ | 170 ± 16 ‡ | 171 ± 82 ‡ |
| Glycated Hb (%) | 4.2 ± 0.8 | 6.9 ± 1 ‡ | 7.3 ± 1 ‡ | 7.5 ± 0.7 ††,‡ |
| Serum Cr (mg/dl) | 0.94 ± 0.13 | 0.96 ± 0.10 | 0.91 ± 0.15 | 0.96 ± 0.23 |
| C-reactive protein (mg/l) | 1.9 (0.2–3.6) | 2 (0.8–4) | 5.1 (1.1–8) **,‡ | 5.8 (3–9.3) **,‡,‡‡ |
| Serum tumour necrosis factor (pg/ml) | 2.3 (0.7–5) | 2.5 (0.7–7) | 6.2 (0.9–12) **, ‡ | 6.4 (0.9–18) ‡, ‡‡ |
| Urinary albumin excretion (mg/day) | 13 (6–26) | 14 (6–27) | 134 (39–288) § | 660 (320–2630) ‡,§,# |
| Urinary NAG (U/g) | 4.3 (0.8–11) | 4.6 (2–7.9) | 6.8 (1.7–18) ††,## | 7.7 (2.3–88.2) §,§§ |
| Urinary tumour necrosis factor (pg/mg) | 4 (0.8–12) | 7 (2–18) §§ | 10.5 (4–20) **,‡ | 18 (9–29) **,‡,‡‡ |
| Controls ( n = 32) | Normoalbuminuria ( n = 25) | Microalbuminuria ( n = 60) | Macroalbuminuria ( n = 75) | |
|---|---|---|---|---|
| Age (years) | 60 ± 8 | 60 ± 6 | 59 ± 8 | 64 ± 9 * |
| Men/women | 17/15 | 13/12 | 29/26 | 36/33 |
| Duration of diabetes (years) | – | 8.3 ± 1.7 | 12.2 ± 3.4 ** | 11.9 ± 3.8 ** |
| BMI (kg/m 2 ) | 25.7 ± 2.3 | 25.3 ± 3.7 | 26.1 ± 2.4 | 25.6 ± 2.54 |
| Systolic BP (mmHg) | 131 ± 4 | 135 ± 3 † | 138 ± 8 † | 141 ± 9 †,†† |
| Diastolic BP (mmHg) | 76 ± 6 | 80 ± 6 † | 85 ± 8 † | 84 ± 9 † |
| Glucose (mg/dl) | 89 ± 7 | 161 ± 13 ‡ | 170 ± 16 ‡ | 171 ± 82 ‡ |
| Glycated Hb (%) | 4.2 ± 0.8 | 6.9 ± 1 ‡ | 7.3 ± 1 ‡ | 7.5 ± 0.7 ††,‡ |
| Serum Cr (mg/dl) | 0.94 ± 0.13 | 0.96 ± 0.10 | 0.91 ± 0.15 | 0.96 ± 0.23 |
| C-reactive protein (mg/l) | 1.9 (0.2–3.6) | 2 (0.8–4) | 5.1 (1.1–8) **,‡ | 5.8 (3–9.3) **,‡,‡‡ |
| Serum tumour necrosis factor (pg/ml) | 2.3 (0.7–5) | 2.5 (0.7–7) | 6.2 (0.9–12) **, ‡ | 6.4 (0.9–18) ‡, ‡‡ |
| Urinary albumin excretion (mg/day) | 13 (6–26) | 14 (6–27) | 134 (39–288) § | 660 (320–2630) ‡,§,# |
| Urinary NAG (U/g) | 4.3 (0.8–11) | 4.6 (2–7.9) | 6.8 (1.7–18) ††,## | 7.7 (2.3–88.2) §,§§ |
| Urinary tumour necrosis factor (pg/mg) | 4 (0.8–12) | 7 (2–18) §§ | 10.5 (4–20) **,‡ | 18 (9–29) **,‡,‡‡ |
* P < 0.01 vs microalbuminuria; ** P < 0.0001 vs normoalbuminuria; †P < 0.01 vs controls; ††P < 0.05 vs normoalbuminuria; ‡P < 0.0001 vs controls; ‡‡P < 0.0001 vs microalbuminuria; §P < 0.001 vs normoalbuminuria; §§P < 0.001 vs controls; #P < 0.001 vs microalbuminuria; ##P < 0.05 vs controls.
General and clinical parameters in healthy controls and type 2 diabetic patients stratified by albuminuria status
| Controls ( n = 32) | Normoalbuminuria ( n = 25) | Microalbuminuria ( n = 60) | Macroalbuminuria ( n = 75) | |
|---|---|---|---|---|
| Age (years) | 60 ± 8 | 60 ± 6 | 59 ± 8 | 64 ± 9 * |
| Men/women | 17/15 | 13/12 | 29/26 | 36/33 |
| Duration of diabetes (years) | – | 8.3 ± 1.7 | 12.2 ± 3.4 ** | 11.9 ± 3.8 ** |
| BMI (kg/m 2 ) | 25.7 ± 2.3 | 25.3 ± 3.7 | 26.1 ± 2.4 | 25.6 ± 2.54 |
| Systolic BP (mmHg) | 131 ± 4 | 135 ± 3 † | 138 ± 8 † | 141 ± 9 †,†† |
| Diastolic BP (mmHg) | 76 ± 6 | 80 ± 6 † | 85 ± 8 † | 84 ± 9 † |
| Glucose (mg/dl) | 89 ± 7 | 161 ± 13 ‡ | 170 ± 16 ‡ | 171 ± 82 ‡ |
| Glycated Hb (%) | 4.2 ± 0.8 | 6.9 ± 1 ‡ | 7.3 ± 1 ‡ | 7.5 ± 0.7 ††,‡ |
| Serum Cr (mg/dl) | 0.94 ± 0.13 | 0.96 ± 0.10 | 0.91 ± 0.15 | 0.96 ± 0.23 |
| C-reactive protein (mg/l) | 1.9 (0.2–3.6) | 2 (0.8–4) | 5.1 (1.1–8) **,‡ | 5.8 (3–9.3) **,‡,‡‡ |
| Serum tumour necrosis factor (pg/ml) | 2.3 (0.7–5) | 2.5 (0.7–7) | 6.2 (0.9–12) **, ‡ | 6.4 (0.9–18) ‡, ‡‡ |
| Urinary albumin excretion (mg/day) | 13 (6–26) | 14 (6–27) | 134 (39–288) § | 660 (320–2630) ‡,§,# |
| Urinary NAG (U/g) | 4.3 (0.8–11) | 4.6 (2–7.9) | 6.8 (1.7–18) ††,## | 7.7 (2.3–88.2) §,§§ |
| Urinary tumour necrosis factor (pg/mg) | 4 (0.8–12) | 7 (2–18) §§ | 10.5 (4–20) **,‡ | 18 (9–29) **,‡,‡‡ |
| Controls ( n = 32) | Normoalbuminuria ( n = 25) | Microalbuminuria ( n = 60) | Macroalbuminuria ( n = 75) | |
|---|---|---|---|---|
| Age (years) | 60 ± 8 | 60 ± 6 | 59 ± 8 | 64 ± 9 * |
| Men/women | 17/15 | 13/12 | 29/26 | 36/33 |
| Duration of diabetes (years) | – | 8.3 ± 1.7 | 12.2 ± 3.4 ** | 11.9 ± 3.8 ** |
| BMI (kg/m 2 ) | 25.7 ± 2.3 | 25.3 ± 3.7 | 26.1 ± 2.4 | 25.6 ± 2.54 |
| Systolic BP (mmHg) | 131 ± 4 | 135 ± 3 † | 138 ± 8 † | 141 ± 9 †,†† |
| Diastolic BP (mmHg) | 76 ± 6 | 80 ± 6 † | 85 ± 8 † | 84 ± 9 † |
| Glucose (mg/dl) | 89 ± 7 | 161 ± 13 ‡ | 170 ± 16 ‡ | 171 ± 82 ‡ |
| Glycated Hb (%) | 4.2 ± 0.8 | 6.9 ± 1 ‡ | 7.3 ± 1 ‡ | 7.5 ± 0.7 ††,‡ |
| Serum Cr (mg/dl) | 0.94 ± 0.13 | 0.96 ± 0.10 | 0.91 ± 0.15 | 0.96 ± 0.23 |
| C-reactive protein (mg/l) | 1.9 (0.2–3.6) | 2 (0.8–4) | 5.1 (1.1–8) **,‡ | 5.8 (3–9.3) **,‡,‡‡ |
| Serum tumour necrosis factor (pg/ml) | 2.3 (0.7–5) | 2.5 (0.7–7) | 6.2 (0.9–12) **, ‡ | 6.4 (0.9–18) ‡, ‡‡ |
| Urinary albumin excretion (mg/day) | 13 (6–26) | 14 (6–27) | 134 (39–288) § | 660 (320–2630) ‡,§,# |
| Urinary NAG (U/g) | 4.3 (0.8–11) | 4.6 (2–7.9) | 6.8 (1.7–18) ††,## | 7.7 (2.3–88.2) §,§§ |
| Urinary tumour necrosis factor (pg/mg) | 4 (0.8–12) | 7 (2–18) §§ | 10.5 (4–20) **,‡ | 18 (9–29) **,‡,‡‡ |
* P < 0.01 vs microalbuminuria; ** P < 0.0001 vs normoalbuminuria; †P < 0.01 vs controls; ††P < 0.05 vs normoalbuminuria; ‡P < 0.0001 vs controls; ‡‡P < 0.0001 vs microalbuminuria; §P < 0.001 vs normoalbuminuria; §§P < 0.001 vs controls; #P < 0.001 vs microalbuminuria; ##P < 0.05 vs controls.
Diabetic patients with normoalbuminuria had higher systolic and diastolic BP as well as higher levels of glucose and HbA1c than healthy non-diabetic individuals ( Table 1 ). Regarding inflammatory parameters, the only difference between both groups was the urinary TNFα excretion, which was significantly higher in normoalbuminuric diabetic patients than in controls: 7 (2–18) vs 4 (0.8–12) pg/mg, respectively ( P < 0.001). When diabetic patients were stratified by albuminuria status ( Table 1 ), a significant difference was observed in serum levels of hs-CRP and TNFα between patients with increased urinary albumin excretion in comparison with subjects with normoalbuminuria. Regarding urinary TNFα, the excretion of this cytokine in subjects with micro- or macroalbuminuria was also significantly higher than in individuals with a normal UAE. A similar finding was observed in patients with macroalbuminuria with respect to subjects with microalbuminuria. Serum and urinary levels of TNFα and hs-CRP in control subjects and diabetic patients are presented in Figure 1 . Patients on the upper quartile for these inflammatory parameters had more albuminuria and UNAG than patients on the lower quartile ( Table 2 ). Thus, UAE and UNAG in patients on the 75th percentile for urinary TNFα excretion were 900 (167–2630) mg/day and 11.2 (2.3–88.2) U/g, respectively, whereas in patients on the 25th percentile albuminuria was 42 (6–220) mg/day and UNAG was 4.6 (1.7–9.2) U/g ( P < 0.001).
Serum concentrations of hs-CRP (mg/l) (white bars), and serum (pg/ml) (grey bars) and urinary (pg/mg) (black bars) levels of TNFα in healthy control individuals and type 2 diabetic patients stratified according albuminuria status: normoalbuminuria (NAB), microalbuminuria (MicrAB) and macroalbuminuria (MacrAB).
Urinary albumin and N-acetyl-β-glucosaminidase (NAG) excretion according to lower (25th) and higher (75th) percentiles for inflammatory parameters in diabetic patients
| 25th percentile | 75th percentile | P -value | |
|---|---|---|---|
| Serum tumour necrosis factor-α | |||
| Urinary albumin excretion (mg/day) | 123 (8–1500) | 660 (50–1400) | <0.0001 |
| Urinary NAG excretion (U/g) | 5.1 (2–21) | 7.8 (1.7–37.7) | <0.001 |
| Urinary tumour necrosis factor-α | |||
| Urinary albumin excretion (mg/day) | 42 (6–220) | 900 (167–2630) | <0.0001 |
| Urinary NAG excretion (U/g) | 4.6 (1.7–9.2) | 11.2 (2.3–88.2) | <0.0001 |
| hs-CRP | |||
| Urinary albumin excretion (mg/day) | 26.5 (6–1500) | 850 (105–2630) | <0.0001 |
| Urinary NAG excretion (U/g) | 5 (2–11.7) | 10.7 (3–88.2) | <0.001 |
| 25th percentile | 75th percentile | P -value | |
|---|---|---|---|
| Serum tumour necrosis factor-α | |||
| Urinary albumin excretion (mg/day) | 123 (8–1500) | 660 (50–1400) | <0.0001 |
| Urinary NAG excretion (U/g) | 5.1 (2–21) | 7.8 (1.7–37.7) | <0.001 |
| Urinary tumour necrosis factor-α | |||
| Urinary albumin excretion (mg/day) | 42 (6–220) | 900 (167–2630) | <0.0001 |
| Urinary NAG excretion (U/g) | 4.6 (1.7–9.2) | 11.2 (2.3–88.2) | <0.0001 |
| hs-CRP | |||
| Urinary albumin excretion (mg/day) | 26.5 (6–1500) | 850 (105–2630) | <0.0001 |
| Urinary NAG excretion (U/g) | 5 (2–11.7) | 10.7 (3–88.2) | <0.001 |
Urinary albumin and N-acetyl-β-glucosaminidase (NAG) excretion according to lower (25th) and higher (75th) percentiles for inflammatory parameters in diabetic patients
| 25th percentile | 75th percentile | P -value | |
|---|---|---|---|
| Serum tumour necrosis factor-α | |||
| Urinary albumin excretion (mg/day) | 123 (8–1500) | 660 (50–1400) | <0.0001 |
| Urinary NAG excretion (U/g) | 5.1 (2–21) | 7.8 (1.7–37.7) | <0.001 |
| Urinary tumour necrosis factor-α | |||
| Urinary albumin excretion (mg/day) | 42 (6–220) | 900 (167–2630) | <0.0001 |
| Urinary NAG excretion (U/g) | 4.6 (1.7–9.2) | 11.2 (2.3–88.2) | <0.0001 |
| hs-CRP | |||
| Urinary albumin excretion (mg/day) | 26.5 (6–1500) | 850 (105–2630) | <0.0001 |
| Urinary NAG excretion (U/g) | 5 (2–11.7) | 10.7 (3–88.2) | <0.001 |
| 25th percentile | 75th percentile | P -value | |
|---|---|---|---|
| Serum tumour necrosis factor-α | |||
| Urinary albumin excretion (mg/day) | 123 (8–1500) | 660 (50–1400) | <0.0001 |
| Urinary NAG excretion (U/g) | 5.1 (2–21) | 7.8 (1.7–37.7) | <0.001 |
| Urinary tumour necrosis factor-α | |||
| Urinary albumin excretion (mg/day) | 42 (6–220) | 900 (167–2630) | <0.0001 |
| Urinary NAG excretion (U/g) | 4.6 (1.7–9.2) | 11.2 (2.3–88.2) | <0.0001 |
| hs-CRP | |||
| Urinary albumin excretion (mg/day) | 26.5 (6–1500) | 850 (105–2630) | <0.0001 |
| Urinary NAG excretion (U/g) | 5 (2–11.7) | 10.7 (3–88.2) | <0.001 |
Unadjusted linear regression analysis showed that UNAG excretion was significant and positively associated with urinary TNFα concentration ( R = 0.52), diabetes duration ( R = 0.51), albuminuria ( R = 0.48), serum levels of hs-CRP ( R = 0.46) and HbA1c ( R = 0.42) ( P < 0.001 in all cases), as well as with serum Cr ( R = 0.27), serum TNFα ( R = 0.18) and SBP ( R = 0.17) ( P < 0.05 in all cases). Regarding UAE, this parameter was significantly correlated with urinary TNFα excretion ( R = 0.64), hs-CRP ( R = 0.52), duration of diabetes ( R = 0.41) and HbA1c ( R = 0.29) ( P < 0.001 in all cases), as well as with serum Cr ( R = 0.25, P < 0.01), serum glucose ( R = 0.21, P < 0.01) and serum TNFα ( R = 0.20, P < 0.05). The associations between urinary TNFα with albuminuria and UNAG remained significant both in patients with microalbuminuria ( R = 0.55 and R = 0.56, respectively, P < 0.001) and in patients with macroalbuminuria ( R = 0.31, P < 0.01 and R = 0.53, P < 0.001, respectively). However, these parameters were not significantly related in subjects with normal UAE. Serum levels of TNFα and hs-CRP were not related to UAE or UNAG excretion in diabetic patients. Likewise, urinary levels of TNFα were not significantly correlated with the serum concentration of this cytokine in any group of diabetic subjects. Finally, serum, but not urinary, TNFα levels were significantly correlated with BMI ( R = 0.16, P < 0.05), whereas only urinary TNFα was significantly associated with HbA1c ( R = 0.46, P < 0.01).
After adjusting for the effect of other variables by partial correlation analysis, the previous associations between urinary TNFα and UAE and UNAG remained significant, but not the relationship between serum TNFα and UNAG ( Table 3 ). Finally, forward stepwise multiple regression analysis was performed to examine the associations between UAE and UNAG (dependent variables) on the one hand and relevant patient characteristics and clinical parameters (i.e. age, gender, time of diabetes, systolic and diastolic BP, serum Cr, HbA1c, hs-CRP and serum and urinary TNFα) as the independent variables on the other. The result showed that urinary TNFα ( P < 0.0001), hs-CRP ( P < 0.0001), serum TNFα ( P < 0.01) and HbA1c ( P < 0.05) were independent of and significantly associated with UAE according to the following function (adjusted R2 = 0.63, P < 0.001): UAE = 0.31 + (0.09 × urinary TNFα) + (0.24 × hs-CRP) + (0.06 × serum TNFα) + (0.17 × HbA1c). Concerning UNAG, duration of diabetes ( P < 0.001), urinary TNFα ( P < 0.01), HbA1c ( P = 0.01), hs-CRP ( P < 0.05) and serum Cr ( P < 0.05) were independent of and significantly associated with this variable according to the following expression (adjusted R2 = 0.44, P < 0.001): UNAG = −28.6 + (0.26 × time of diabetes) + (0.23 × urinary TNFα) + (0.17 × HbA1c) + (0.17 × hs-CRP) + (0.15 × serum Cr).
Significant associations between the independent variables albuminuria and urinary NAG excretion and other parameters by partial correlation analysis
| Variables | Correlation coefficient | P -value |
|---|---|---|
| Urinary albumin excretion | ||
| Urinary tumour necrosis factor-α | R = 0.49 | <0.0001 |
| C-reactive protein | R = 0.44 | <0.0001 |
| Serum tumour necrosis factor-α | R = 0.20 | <0.01 |
| Glycated haemoglobin | R = 0.16 | <0.05 |
| Urinary NAG excretion | ||
| Duration of diabetes | R = 0.29 | <0.001 |
| Urinary tumour necrosis factor-α | R = 0.24 | <0.01 |
| Glycated haemoglobin | R = 0.19 | 0.01 |
| C-reactive protein | R = 0.18 | <0.05 |
| Serum creatinine | R = 0.18 | <0.05 |
| Variables | Correlation coefficient | P -value |
|---|---|---|
| Urinary albumin excretion | ||
| Urinary tumour necrosis factor-α | R = 0.49 | <0.0001 |
| C-reactive protein | R = 0.44 | <0.0001 |
| Serum tumour necrosis factor-α | R = 0.20 | <0.01 |
| Glycated haemoglobin | R = 0.16 | <0.05 |
| Urinary NAG excretion | ||
| Duration of diabetes | R = 0.29 | <0.001 |
| Urinary tumour necrosis factor-α | R = 0.24 | <0.01 |
| Glycated haemoglobin | R = 0.19 | 0.01 |
| C-reactive protein | R = 0.18 | <0.05 |
| Serum creatinine | R = 0.18 | <0.05 |
Significant associations between the independent variables albuminuria and urinary NAG excretion and other parameters by partial correlation analysis
| Variables | Correlation coefficient | P -value |
|---|---|---|
| Urinary albumin excretion | ||
| Urinary tumour necrosis factor-α | R = 0.49 | <0.0001 |
| C-reactive protein | R = 0.44 | <0.0001 |
| Serum tumour necrosis factor-α | R = 0.20 | <0.01 |
| Glycated haemoglobin | R = 0.16 | <0.05 |
| Urinary NAG excretion | ||
| Duration of diabetes | R = 0.29 | <0.001 |
| Urinary tumour necrosis factor-α | R = 0.24 | <0.01 |
| Glycated haemoglobin | R = 0.19 | 0.01 |
| C-reactive protein | R = 0.18 | <0.05 |
| Serum creatinine | R = 0.18 | <0.05 |
| Variables | Correlation coefficient | P -value |
|---|---|---|
| Urinary albumin excretion | ||
| Urinary tumour necrosis factor-α | R = 0.49 | <0.0001 |
| C-reactive protein | R = 0.44 | <0.0001 |
| Serum tumour necrosis factor-α | R = 0.20 | <0.01 |
| Glycated haemoglobin | R = 0.16 | <0.05 |
| Urinary NAG excretion | ||
| Duration of diabetes | R = 0.29 | <0.001 |
| Urinary tumour necrosis factor-α | R = 0.24 | <0.01 |
| Glycated haemoglobin | R = 0.19 | 0.01 |
| C-reactive protein | R = 0.18 | <0.05 |
| Serum creatinine | R = 0.18 | <0.05 |
Discussion
In the present study, we analysed the serum levels and urinary TNFα excretion in type 2 diabetic patients and healthy non-diabetic controls. Our results show that serum and urinary concentrations of this cytokine are enhanced in diabetic subjects, increasing as renal damage progresses. More importantly urinary TNFα is significant and independently related to clinical markers of both glomerular (UAE: R = 0.49, P < 0.0001) and tubulointerstitial (UNAG: R = 0.24, P < 0.01) injury. Works in animal models as well as other smaller clinical studies have reported interesting data about the importance of TNFα in the setting of DN. Our results highlight the possible role of TNFα, one of the main pro-inflammatory cytokines, in the development and progression of renal injury in diabetic patients.
Serum TNFα has been reported to be elevated in diabetic subjects, and even in patients with only impaired glucose tolerance, compared to healthy individuals [ 11 , 12 ]. In the present work, diabetic patients had approximately a 2.5-fold higher serum TNFα than non-diabetic individuals. However, there were no significant differences in serum TNFα between diabetic subjects with normoalbuminuria and non-diabetic controls. In fact, serum TNFα concentrations were increased only in diabetic patients with micro- or macroalbuminuria. Previous works have reported similar results, and furthermore, they showed a significant relationship between serum TNFα and UAE [ 5 , 13 , 14 ]. In our work, after multivariate adjusted analysis, serum TNFα was significantly related to UAE, but not to UNAG.
In the present study, there were no significant differences in BMI among diabetic patients after stratification by albuminuria status, whereas diabetic subjects with micro- or macroalbuminuria had higher levels of HbA1c than patients with normal UAE. After regression analysis, serum but not urinary TNFα levels were significantly correlated with BMI, whereas only urinary TNFα was significantly associated with HbA1c. These findings are in agreement with the results of previous studies, and suggest that serum levels of TNFα are essentially determined by the systemic production of this adipocytokine by adipose tissue, unrelated to metabolic factors such as HbA1c [ 15 , 16 ]. On the contrary, urinary concentration of TNFα represents the local production of this cytokine within the kidney, which would be significantly related to metabolic control [ 17 ].
The local actions and implications of intrarenal TNFα are extremely interesting. In 1991, Hasegawa et al . [ 4 ] reported that macrophages incubated with glomerular basement membranes from diabetic rats produced greater amounts of TNFα than did macrophages incubated with membranes from non-diabetic animals. These authors suggested for the first time the potential implication of this cytokine in the pathogenesis of DN. TNFα is a pleiotropic cytokine playing an essential role in the molecular basis of inflammatory and immunological reactions. Furthermore, TNFα is a potential causative agent implicated in glomerular and tubulointerstitial damage in the setting of diabetes.
Recent experimental works have demonstrated that TNFα gene expression is increased in diabetic rats as compared to control rats, both in glomerular and tubulointerstitial structures [ 10 , 18–20 ]. It is important to note that this cytokine may be produced in the diabetic kidney by infiltrating cells (mainly macrophages) as well as intrinsically by renal cells (endothelial, mesangial, glomerular and tubular epithelial cells) [ 21 , 22 ]. Within the glomerulus, TNFα displays a number of effects that are relevant to the manifestations observed at the initial stages and during the progression of glomerular injury. TNFα is implicated in the disbalance between vasodilatory and vasoconstrictive mediators, which may result in alterations of glomerular blood flow and glomerular filtration rate [ 23 ]. Additionally, this cytokine is cytotoxic to glomerular, mesangial and epithelial cells, and may induce direct renal damage [ 8 , 24 ]. TNFα is able to promote the local generation of reactive oxygen species, which affects the barrier function of the glomerular capillary wall resulting in enhanced albumin permeability, independently of haemodynamic factors or effects of recruited inflammatory cells [ 9 ]. Studies in diabetic rats have demonstrated that UAE significantly correlates with renal cortical mRNA levels and urinary TNFα excretion, and moreover, increased TNFα concentrations in urine as well as in renal interstitial fluid preceded the rise in UAE [ 10 , 19 ]. Regarding tubulointerstitial damage, exposure of tubular epithelial cells to TNFα increased the synthesis and secretion of lymphocyte and neutrophil chemoattractant factors [ 25 ] as well as the cell surface expression of intercellular adhesion molecule-1 [ 26 , 27 ], which has been implicated in the development of renal injury in diabetes [ 27 ]. Finally, TNFα has stimulatory effects on sodium uptake by proximal tubule cells [ 28 ], contributing to sodium retention and renal hypertrophy, typical alterations that occur during the early stage of DN [ 20 ]. The results of these studies, as well as the findings in the present work, suggest that urinary TNFα may reflect the intrarenal activity of this cytokine, which may be related to the development and progression of renal injury in diabetes.
An interesting finding in our study was the lack of correlation between urinary TNFα with UAE and UNAG in normoalbuminuric diabetic patients, whereas it is significantly associated with these parameters in diabetic subjects with increased UAE. These findings may suggest two possibilities: (i) the presence of a threshold phenomenon, and therefore, it is very interesting to speculate that by maintaining the urinary TNFα concentration below a certain level it will be possible to offer significant renoprotection; and (ii) the importance of diabetes duration: in our study, albuminuric diabetic patients had a higher duration of diabetes than normoalbuminuric subjects. Therefore, it is possible to hypothesize that the higher time of diabetes, the higher intrarenal effect of several local factors, such as TNF-α, which may contribute significantly to the magnitude of renal injury.
Clinical studies analysing the urinary excretion of TNFα in diabetic patients are scarce. Two previous works measured this parameter in the setting of DN, although the main objective was to study patients with primary glomerulonephritis. In the first work, TNFα was measured in 6 diabetic subjects, with urinary TNFα activity being present only in 1 patient [ 29 ]. In the second study [ 30 ], the median urinary TNFα in 16 patients with DN was 14 (range 8–52) pg/ml, similar to the values found by us in the present work. Finally, in a preliminary study by our group in 65 diabetic patients [ 5 ], urinary TNFα was significantly correlated with UAE. The present study confirms these preliminary findings in a large group of diabetic patients; in addition, it shows that urinary TNFα excretion is also significant, direct and independently related to UNAG, a clinical marker of tubulointerstitial damage.
The importance of this association is highlighted by the effects of therapeutic interventions based on the modulation of TNFα. Pentoxifylline (PTF) is a phosphodiesterase inhibitor with significant immunoregulatory and anti-inflammatory properties. PTF inhibits the accumulation of TNFα mRNA and the transcription of the TNFα gene, suppressing the synthesis and production of this cytokine [ 31 ]. Recent studies by DiPetrillo et al . [ 32 ] and by our group (our unpublished data) in experimental DN have found that PTF administration was able to prevent the increased renal TNFα expression, synthesis and excretion during diabetes. In addition, inhibition of TNFα by PTF was associated with a reduction in sodium retention and renal hypertrophy. Together with these experimental observations, clinical trials have shown that PTF reduces urinary protein and NAG excretion in diabetic subjects, both with normal renal function and renal insufficiency [ 13 , 33–35 ]. The antiproteinuric action of PTF was not related to an improvement of metabolic or haemodynamic factors, but interestingly, it was direct and significantly related to a modulation of TNFα activity [ 13 , 35 ].
Finally, our study may have some potential limitations. First, the differences in several parameters among patients after stratification by albuminuria status. However, these are the actual characteristics of diabetic subjects that attend to out-patient clinics daily. Second, the influence of the blockade of renin–angiotensin system and BP on serum TNFα. In our study, since ACEI and ARA are able to reduce TNFα levels, all patients were treated with the maximum recommended doses of these drugs in order to avoid this confounding factor derived from different dosage. On the other hand, there was no significant association between systolic or diastolic BP with TNFα levels. And third, it is possible to consider that the correlation between urinary TNFα and albuminuria exists due to broader glomerular damage reflected by UAE. However, an important finding in our study is that urinary TNFα was not correlated with the serum levels of this cytokine. This strongly suggests that intrarenal TNFα production is involved in urinary TNFα excretion, which is not linked to glomerular damage. In addition, recent studies in experimental models of diabetic nephropathy have demonstrated that renal TNFα gene expression was increased in diabetic rats, with a significant correlation between renal TNFα expression and urinary TNFα levels [ 10 ]. Moreover, it has been shown that increased TNFα concentrations in urine as well as in renal interstitial fluid preceded the rise in UAE [ 16 ].
In conclusion, the findings of the present study show that in type 2 diabetic patients, urinary TNFα excretion is elevated and correlates with severity of renal disease in terms of both glomerular and tubulointerstitial damage. This suggests that TNFα might play a significant role in the pathogenesis and progression of renal injury in diabetes. Prognostic clinical studies will be necessary to evaluate urinary TNFα excretion as a parameter for monitoring the development and evolution of DN. In addition, the potential relevance of these findings with respect to new preventive and therapeutic strategies in the setting of DN deserves future investigation.
Acknowledgments
This study was supported by Fundación Canaria de Investigación y Salud (FUNCIS) and Asociación Científica para la Investigación Nefrológica (ACINEF).
Conflict of interest statement . None declared.
Comments