Connective Tissue Growth Factor Is Increased in Plasma of Type 1 Diabetic Patients with Nephropathy

Pathophysiology/Complications ORIGINAL ARTICLE

Connective Tissue Growth Factor Is Increased in Plasma of Type 1 Diabetic Patients With Nephropathy

1 3 PEGGY ROESTENBERG, MSC WILMAR M. WIERSINGA, MD, PHD these diabetes complications. One of the 1 5 FRANS A. VAN NIEUWENHOVEN, PHD NOELYNN OLIVER, PHD important growth factors involved in 1 5 LOTTE WIETEN, MSC WILLIAM USINGER, PHD 2 5 ECM accumulation and ﬁbrotic processes PETER BOER, PHD STEPHEN WEITZ, PHD ␤ 3 4 is transforming growth factor (TGF)- THEO DIEKMAN, MD, PHD REINIER O. SCHLINGEMANN, MD, PHD 4 1 (1). In both type 1 and type 2 diabetes, ANNA M. TILLER, MD ROEL GOLDSCHMEDING, MD, PHD plasma levels of TGF-␤ were correlated with the presence of diabetes complications (5,6). Connective tissue growth factor OBJECTIVE — Connective tissue growth factor (CTGF) is strongly upregulated in ﬁbrotic (CTGF) is another important growth fac- disorders and has been hypothesized to play a role in the development and progression of diabetes complications. The aim of the present study was to investigate the possible association tor implicated in the development of dia- of plasma CTGF levels in type 1 diabetic patients with markers relevant to development of betes complications. CTGF is a 38-kDa, diabetes complications. cystein-rich secreted protein that was originally cloned from human umbilical

RESEARCH DESIGN AND METHODS — Plasma CTGF levels (full-length and NH2- vein endothelial cells (7). Strong CTGF terminal fragments) were determined in 62 well-characterized patients with type 1 diabetes and expression has been reported (8,9) in ath- in 21 healthy control subjects. Correlations of these plasma CTGF levels with markers of glyce- erosclerotic aorta and in renal mesangial mic control, platelet activation, endothelial activation, nephropathy, and retinopathy were in- cells cultured in high-glucose medium. vestigated. CTGF was also shown (10–12) to be strongly expressed in glomeruli of dia- RESULTS — Elevated plasma NH2-terminal fragment of CTGF (CTGF-N) levels were detected in a subpopulation of type 1 diabetic patients and were associated with diabetic nephrop- betic patients and animals with nephropathy. Stepwise regression analysis revealed contribution of albuminuria, creatinine clearance, athy. Similarly, the magnitude of urinary and duration of diabetes as predictors of plasma CTGF-N level. Elevation of plasma CTGF-N CTGF excretion was related to the sever- levels in patients with retinopathy was probably due to renal comorbidity. ity of diabetic nephropathy (DN) in a cross-sectional study (13) of patients with CONCLUSIONS — Plasma CTGF-N levels are elevated in type 1 diabetic patients with type 1 diabetes. Both high glucose con- nephropathy and appear to be correlated with proteinuria and creatinine clearance. Further centrations and advanced glycation end studies will be needed to determine the relevance of plasma CTGF as a clinical marker and/or products (AGEs) are able to induce ECM pathogenic factor in diabetic nephropathy. production via CTGF (9,14). CTGF itself Diabetes Care 27:1164–1170, 2004 can act as a downstream mediator of TGF-␤ in ECM synthesis, but TGF-␤– independent regulation of CTGF has also been reported (9,15). Therefore, the anal- iabetic patients frequently develop macro- and microvascular damage, extra- ysis of CTGF plasma levels of diabetic pa- severe chronic complications like cellular matrix (ECM) accumulation, and tients might provide important additional D cardiovascular disease, nephropa- eventually chronic fibrosis. It has been information about the involvement of this thy, neuropathy, and retinopathy. Char- shown (1–4) that growth factors play an growth factor in the development of dia- acteristics of these complications are important role in the development of ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● betes complications. Different fragments of the CTGF protein have been detected From the 1Department of Pathology, University Medical Center Utrecht, Utrecht, the Netherlands; the in vitro and in vivo, and at least some of 2Department of Nephrology and Hypertension, University Medical Center Utrecht, Utrecht, the Nether- lands; the 3Department of Endocrinology & Metabolism, Academic Medical Center, Amsterdam, the Neth- these have biological activity. The CTGF erlands; the 4Department of Ophthalmology, Academic Medical Center, Amsterdam, the Netherlands; and molecule, which consists of four mod- 5FibroGen, South San Francisco, California. ules, is mostly cleaved between modules Address correspondence and reprint requests to Dr. Roel Goldschmeding, University Medical Center, II and III, yielding fragments of 16–20 Department of Pathology, H04.312, Heidelberglaan 100, 3584 CX Utrecht, the Netherlands. E-mail: kDa, but smaller fragments have also been [email protected]. Received for publication 4 September 2003 and accepted in revised form 29 January 2004. identified (16,17). The relative contribu- R.O.S. has received research funds from FibroGen. R.G. has received research funds and fees for speaking tion to fibrotic processes of full-length engagements from FibroGen. compared with fragmented CTGF re- Abbreviations: AGE, advanced glycation end product; BMP, bone morphogenetic protein; CTGF, con- mains to be established. nective tissue growth factor; CTGF-N, NH2-terminal fragment of CTGF; DN, diabetic nephropathy; ECM, extracellular matrix; ELISA, enzyme-linked immunosorbent assay; TGF, transforming growth factor; vWF, The aim of the present study was von Willebrand factor. to investigate the possible association © 2004 by the American Diabetes Association. of plasma CTGF levels in 62 well-

1164 DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004 Roestenberg and Associates

characterized type 1 diabetic patients performance liquid chromatography. CTGF ELISA, the capture antibody binds with DN and diabetic retinopathy as well Plasma creatinine concentrations were the COOH-terminal part of the CTGF as with general patient characteristics and determined by automated spectrophoto- protein, whereas the detecting antibody other markers relevant to the develop- metrical assay using Creatinine-PAP (per- (the same as in the CTGF-N ELISA) binds ment of diabetes complications, i.e., gly- oxidase-antiperoxidase). In 24-h urine the NH2-terminal part of the CTGF pro- cemic control, endothelial activation, and samples, creatinine was measured using tein. To avoid confusion due to differ- platelet activation (2,18–20). the Jaffe´ method, albumin was deter- ences in the molecular weight of full- mined with an immunonephelometric as- length and different fragments of CTGF, say, and renal creatinine clearance was all levels were expressed as picomoles per RESEARCH DESIGN AND calculated from these data. liter. The detection limit of these assays METHODS — For the present study, was 4 pmol/l for CTGF-N and 8 pmol/l for we analyzed plasma samples obtained Measurement of markers for platelet full-length CTGF, and intra- and interas- from 62 well-characterized patients with and endothelial activation say variations were 6 and 20%, respec- type 1 diabetes and 21 healthy control Von Willebrand factor (vWF) antigen was tively. subjects. All gave their informed consent determined by sandwich enzyme-linked TGF-␤1 was determined in PECT before participation in the study. The immunosorbent assay (ELISA) using rab- plasma by means of sandwich ELISA ac- study protocol has been approved by the bit anti-human vWF as “capture” anti- cording to the manufacturer’s protocol local ethics committee/institutional board body and horseradish peroxidase– (R&D Systems, Minneapolis, MN). and was conducted according to the prin- conjugated rabbit anti-human vWF as ciples of the Declaration of Helsinki. After “detecting” antibody (Dakopatts, Statistical analysis an overnight fast and abstaining from vig- Glostrup, Denmark). ␤-Thromboglobu- Data are expressed as mean Ϯ SD. Mann- orous physical activity during the previ- lin, a marker for in vivo platelet activa- Whitney analysis and Kruskal-Wallis ous 24 h, patients and control subjects tion, and platelet factor 4, a marker for ex analysis followed by Dunn’s method were presented at the outpatient clinic between vivo platelet activation, were determined performed to determine differences in 8:00 and 10:00 A.M., bringing their 24-h in platelet-depleted PECT plasma by plasma CTGF-N levels between groups. urine. Demographic and relevant medical means of sandwich ELISAs according to Forward stepwise regression analysis was history data were recorded, including age, the manufacturer’s instructions (Assera- used to compare CTGF-N levels with gen- sex, duration of diabetes, insulin dose, co- chrom; Diagnostica Stago) (21). eral patient characteristics, glycemic con- morbidity, medication, and smoking hab- trol, endothelial activation, platelet its. Blood pressure was measured with a ELISAs for plasma CTGF (NH2-terminal activation, albuminuria, and creatinine sphygmomanometer in the sitting position; and full length) and TGF-␤ clearance. Because data were not nor- the median of three successive measure- Plasma content of full-length and N- mally distributed, Spearman’s correla- ments was noted. Height and weight were terminal fragments (CTGF-N) of CTGF tions were calculated between CTGF-N measured to determine BMI. Retinopathy were determined by means of two sepa- levels and HbA1c, albuminuria, and was scored by a single experienced ophthal- rate sandwich ELISAs, each using two dis- TGF-␤1 levels. In all cases, P Ͻ 0.05 was mologist by fundoscopic examination and tinct monoclonal antibodies against considered significant. examination of the clinical charts. human CTGF (FibroGen, South San Blood was collected by puncture of an Francisco, CA). Microtiter plates were RESULTS antebrachial vein using sodium heparin, coated overnight at 4°C with capture an- citrate, or PECT as anticoagulants, depen- tibody and blocked with BSA. Citrate Plasma CTGF-N level is elevated in dent on the assay. PECT medium (400 plasma was diluted 5- or 10-fold in assay DN ␮ l/4.5 ml polypropylene tube), used to buffer, and a 50-␮l diluted sample was Levels of full-length CTGF were below prevent ex vivo platelet activation, is com- added to each well together with 50 ␮l the sensitivity limit of our sandwich posed of equal volumes of solutions A, B, biotinylated CTGF detection antibody. ELISA in all control as well as diabetic ϭ and C, in which A 282 nmol/l prosta- After incubation for2hat37°C followed plasma samples tested, although full- ϭ glandin E1 and 1.9 mmol/l Na2CO3,B by incubation with streptavidin- length recombinant CTGF was readily de- ϭ 30 mmol/l theophylline in PBS, and C conjugated alkaline phosphatase for1hat tectable if spiked in these plasma samples. 270 mmol/l EDTA. To further avoid arti- room temperature, plates were washed This means that these plasma samples ficial platelet and leukocyte activation, and 100 ␮l of substrate solution contain- contained Ͻ80 pmol/l of full-length blood samples were immediately cooled ing p-nitrophenyl phosphate was added CTGF (not shown). In contrast, CTGF-N on ice and platelets were depleted by cen- to each well. After 20 min of color devel- was readily detectable in the majority of trifuging the samples for 60 min at 4,000g opment, absorbance was read at 405 nm. these same plasma samples. No signifi- and 4°C within 1 h after collection. All Purified recombinant human CTGF (Fi- cant difference was found in CTGF-N lev- Ϫ plasma samples were stored at 70°C un- broGen) was used for calibration. Both els between control subjects and the total til analysis. monoclonal antibodies used in the group of diabetic patients, but the varia- CTGF-N sandwich ELISA specifically tion was much larger among samples Measurement of markers for bind distinct epitopes on the N-terminal from diabetic patients (138 Ϯ 136 pmol/l glycemic control and nephropathy half of the CTGF protein. This assay de- in diabetic patients vs. 103 Ϯ 51 pmol/l in Concentrations of HbA1c were deter- tects both CTGF-N as well as the full- control subjects) (Fig. 1A). Patients were mined in citrate plasma by means of high- length CTGF protein. In the full-length divided into subgroups according to the

DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004 1165 Elevated plasma CTGF in diabetic nephropathy level of albuminuria, i.e., normoalbumin- minuria or creatinine clearance was different from those of normoalbumin- uria (Ͻ30 mg/day), microalbuminuria taken as a dependent variable in the for- uric patients without retinopathy (n ϭ 17). (30–300 mg/day), and overt proteinuria ward stepwise regression analysis of this (i.e., DN, Ͼ300 mg/day). This revealed dataset, plasma CTGF-N level was iden- CONCLUSIONS — Changes in significant differences in plasma CTGF-N tified as the strongest independent growth factor balance are important in the levels between DN patients and control predictor (R ϭ 0.572 and 0.514, respec- development of chronic complications of subjects and between DN and nor- tively). It thus appears that plasma diabetes. In this study, we observed that moalbuminuric patients (Fig. 1B). Fur- CTGF-N levels are correlated with mark- plasma CTGF-N levels are significantly el- thermore, a very wide variation in plasma ers for nephropathy. evated in patients with DN and correlated CTGF-N levels was noted within the sub- with markers for DN and glycemic con- group of microalbuminuric patients trol. Full-length CTGF levels were below Ϯ Plasma TGF-␤1 levels (203 203 pmol/l). ␤ the detection limit of our ELISA in all di- Since TGF- 1, unlike CTGF (F.A.V.N., abetic and normal plasma samples tested. P.R., unpublished observations) is abun- This absence of detectable full-length Plasma CTGF-N level correlates with dant in platelets and released during CTGF in plasma might be related to tech- HbA1c platelet activation, reliable estimates of “in ␤ nical limitations (sensitivity of the full- General patient characteristics and mark- vivo” plasma TGF- 1 levels can be ob- length ELISA) or to its clearance via ers of glycemic control, endothelial activa- tained only in the absence of significant ex COOH-terminal interaction with matrix tion, platelet activation, and nephropathy of vivo platelet activation (i.e., platelet factor Ͻ components and (scavenging) receptors healthy subjects and normoalbuminuric, 4 10 units/l). Despite careful plasma and to proteolysis, e.g., by matrix metal- microalbuminuric, and DN diabetic pa- collection and processing, only 21 dia- loproteases, plasmin, and elastase, which tients are summarized in Table 1. betic and 5 control samples met this cri- ␤ have all been reported to cleave CTGF HbA1c levels in the diabetic patients terion. In these 26 samples, TGF- 1 (16). correlated, albeit only weakly, with levels did correlate with CTGF-N levels It has been reported (10–13,22) that plasma CTGF-N (R ϭ 0.355, P ϭ 0.005) (R ϭ 0.671, P Ͻ 0.001), but no significant ␤ CTGF mRNA and protein levels are sig- (Fig. 1C). This is in line with induction of difference in plasma TGF- 1 levels was nificantly increased in kidney tissue and CTGF expression by high ambient glu- observed between healthy subjects and urine of patients as well as experimental cose and AGEs observed previously diabetic patients. Within the group of 21 animals with DN. Adler et al. (23) found in vitro (9,14). No significant correlation diabetic patients, TGF-␤1 correlated with ϭ ϭ equally elevated CTGF mRNA levels in between plasma CTGF-N and HbA1c HbA1c (R 0.477, P 0.029). This is glomeruli of microalbuminuric (n ϭ 5) was found within any of the sub- stronger than the correlation between and DN (n ϭ 6) diabetic patients. We now groups. This might be due to the relatively plasma CTGF-N and HbA1c. add to this notion that CTGF-N levels are small numbers of microalbuminuric and In subgroups defined by degree of al- elevated in plasma of almost all type 1 DN patients. buminuria, no significant difference was ␤ diabetic patients with DN and also in found between TGF- 1 levels, although about one-third of microalbuminuric pa- Plasma CTGF-N correlates with there seemed to be a trend to higher levels tients, but only in a small minority of nor- albuminuria and creatinine in patients with DN as compared with di- ϭ moalbuminuric patients. Plasma CTGF-N clearance abetic patients without DN (P 0.114) levels of patients with DN as a group dif- Forward stepwise regression analysis (data not shown). Due to small numbers, ␤ fered significantly from those of nor- showed that albuminuria, creatinine it was not possible to include TGF- 1asa moalbuminuric patients and healthy clearance, and duration of diabetes are parameter in forward stepwise regression control subjects. Due to very wide scatter, independent predictors of plasma analysis. levels in microalbuminuric patients were CTGF-N level. Plasma CTGF-N most not significantly different from those in strongly correlated with albuminuria Plasma CTGF-N levels in patients healthy control subjects or normoalbu- (R ϭ 0.572, P Ͻ 0.001) (Fig. 1D), with retinopathy minuric patients and also not significantly whereas albuminuria and creatinine Diabetic retinopathy is almost invariably different from levels in patients with DN. clearance together yielded a cumulative present in patients with DN. Plasma Since there was a difference in age be- R of 0.667, with P Ͻ 0.001. Correlation CTGF-N levels appeared to be signifi- tween healthy control subjects and mi- of CTGF-N with albuminuria, creati- cantly elevated in patients with retinopa- croalbuminuric and DN diabetic patients, nine clearance, and duration of the dia- thy compared with those without we also measured plasma CTGF-N in an betes together resulted in a cumulative retinopathy (P Ͻ 0.001) (Fig. 1E). How- additional group of 20 healthy subjects R of 0.759, with a P ϭ 0.001. Addition ever, of the 17 patients with diabetic ret- whose mean age (42 Ϯ 8 years) was com- of further parameters did not signifi- inopathy who had Ͼ180 pmol/l of parable with that of the studied mi- cantly contribute to this correlation. CTGF-N in their plasma (upper limit of croalbuminuric and DN diabetic patients. Within the group of DN patients, the normal control subjects), only 4 were In this additional control group, we found plasma CTGF-N levels showed a ten- normoalbuminuric, whereas 4 of the re- a mean plasma CTGF-N level of 120 Ϯ 72 dency to correlate with albuminuria, maining 13 were microalbuminuric and 9 pmol/l, which is in the same range as the but this was not statistically significant proteinuric (i.e., had DN). Plasma plasma CTGF-N levels in the other (n ϭ 10, R ϭ 0.553, P ϭ 0.097). When, CTGF-N levels of normoalbuminuric pa- healthy control group and the nor- instead of plasma CTGF-N level, albu- tients with retinopathy (n ϭ 23) were not moalbuminuric diabetic patient group. It

1166 DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004 Roestenberg and Associates

Figure 1—Plasma CTGF-N levels and correlations in control subjects and different subgroups of type 1 diabetic patients. A: Plasma CTGF-N levels in healthy control subjects (103 Ϯ 51 pmol/l [mean Ϯ SD]) and all diabetic patients (138 Ϯ 136 pmol/l) (P ϭ 0.765). B: Distribution of plasma CTGF-N fragment levels in the different patient subgroups according to albuminuria: healthy control subjects, 103 Ϯ 51 pmol/l; normoalbuminuric (NA), 80 Ϯ 66; microalbuminuric (MA), 203 Ϯ 203; and DN, 290 Ϯ 101 (mean Ϯ SD). A signiﬁcant difference (P Ͻ 0.05) was observed between control subjects and DN patients and between normoalbuminuric and DN patients. C: Correlation between plasma CTGF-N levels and HbA1c in all diabetic patients. Spearman’s correlation: R ϭ 0.355, P ϭ 0.005. D: Correlations between plasma CTGF-N and albuminuria in the total group of type 1 diabetic patients and in the subgroup of DN patients only. Diabetic patients (n ϭ 62): R ϭ 0.572, P Ͻ 0.001. DN patients only (Spearman’s correlation, n ϭ 10): R ϭ 0.553, P ϭ 0.097. E: Distribution of plasma CTGF-N levels in diabetic patients with and without retinopathy: no retinopathy, 61 Ϯ 57 pmol/l, and retinopathy, 178 Ϯ 148 (mean Ϯ SD) (P Ͻ 0.001). Elevated plasma CTGF-N levels are almost exclusively found in microalbuminuric and DN patients. E, normoalbuminuric; ϫ, microalbuminuric; F, DN.

DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004 1167 Elevated plasma CTGF in diabetic nephropathy

Table 1 —General and clinical parameters of healthy control subjects and type 1 diabetic cloned from endothelial cells and athero- patients sclerotic aorta (7,8), markers for endothelial or platelet activation (vWF and Type 1 diabetic patients ␤ Control -thromboglobulin) did not contribute to subjects NA MA DN prediction of plasma CTGF-N levels. Plasma CTGF-N levels were significantly General patient characteristics elevated in the group of diabetic patients n (men) 21 (11) 40 (19) 12 (6) 10 (6) with retinopathy compared with patients Age (years) 30 Ϯ 637Ϯ 10 45 Ϯ 943Ϯ 10 without retinopathy, but patients with Duration of diabetes (years) — 18.3 Ϯ 12.0 23.0 Ϯ 8.8 26.2 Ϯ 6.2 retinopathy who had no signs of renal BMI (kg/m2) 23.4 Ϯ 3.3 23.2 Ϯ 2.7 22.0 Ϯ 2.0 22.6 Ϯ 2.1 complications did not have elevated Systolic blood pressure (mmHg) 129 Ϯ 15 132 Ϯ 18 141 Ϯ 18 152 Ϯ 22 plasma CTGF-N levels. Therefore, the el- Glycemic control evated plasma CTGF-N levels in the Ϯ Ϯ Ϯ Ϯ HbA1c (%) 5.2 0.4 8.0 1.7 8.3 1.1 9.2 1.2 group of retinopathy patients are likely Platelet activation due to associated nephropathy rather ␤-Thromboglobulin (units/l) 28.5 Ϯ 8.8 27.8 Ϯ 6.4 33.4 Ϯ 6.5 34.6 Ϯ 9.9 than to the retinopathy itself. However, Endothelial activation this does not necessarily mean that CTGF vWF (%) 78 Ϯ 18 103 Ϯ 31 111 Ϯ 23 129 Ϯ 43 is more important in nephropathy than in Nephropathy retinopathy. In retinal endothelial cells Albuminuria (mg/24 h) 9.2 Ϯ 7.4 8.9 Ϯ 5.1 102 Ϯ 82 953 Ϯ 690 and in pericytes, CTGF is induced by vas- Creatinine clearance (ml/min) 137 Ϯ 22 131 Ϯ 37 120 Ϯ 46 87 Ϯ 34 cular endothelial growth factor (25). We Data are means Ϯ SD. MA, microalbuminuric; NA, normoalbuminuric. have observed altered CTGF distribution in retinas of diabetic patients with increased positivity of pericytes, which was not possible to include these addi- main characteristics of DN, and a negative might relate to capillary basement mem- tional 20 control subjects in the study but almost equally strong correlation was brane thickening characteristic of diabetic because other parameters were not deter- found with creatinine clearance. Identical retinopathy (E. Kuiper, R.O.S., R.G., un- mined in these individuals. results were obtained when the MDRD published results). Based on the size and The apparent discrepancy with the (modification of diet in renal disease) for- perfusion of the organs, however, local similarly elevated kidney mRNA levels in mula and Cockcroft-Gault calculations production of CTGF in the kidney can be all five microalbuminuric patients and six were used as estimates of glomerular fil- expected to contribute more to plasma patients with DN reported by Adler et al. tration rate instead of creatinine clear- levels than local production in the retina. (23) might indicate that plasma CTGF ance. CTGF (36–38 kDa), and even more The observed correlation of plasma level is not solely determined by renal its fragments (ϳ10–20 kDa), can be ex- CTGF levels with markers for nephropa- CTGF expression. Moreover, microalbu- pected to pass the glomerular filter into thy raises questions as to the relevance of minuric patients are heterogeneous with the primary urine, although dimerization known determinants of the development respect to progression. A recent meta- and other protein interactions might pre- of chronic diabetes complications to reg- analysis by Caramori et al. (24) revealed a vent filtration in normal glomeruli. In ulation of CTGF expression. It is known 30– 45% risk of progression of mi- DN, proteinuria is accompanied by a pro- that CTGF and its major inducer, TGF-␤, croalbuminuria to proteinuria over 10 gressive decline of glomerular filtration. can be both induced by elevated levels of years compared with 30% remission to a Therefore, elevated plasma CTGF-N lev- glucose and AGEs (9,14) and that strict normoalbuminuric state and stabilization els in patients with DN might (at least in glycemic control is important in the pre- of microalbuminuria in the remaining pa- part) reflect loss of renal clearance. In vention of development and progression tients. In this respect, it is of interest that conjunction with albuminuria and creat- of diabetes complications (2). Therefore, our microalbuminuric group showed a inine clearance, duration of diabetes was a correlation might be expected between ␤ remarkable interindividual variation in identified as a third factor that signifi- CTGF, TGF- 1, and HbA1c levels in dia- plasma CTGF-N levels, with significantly cantly contributed to plasma CTGF-N betic plasma. In our study, we indeed elevated levels in 4 of 12 microalbumin- levels. No colinearity between albumin- found a correlation between plasma uric patients. Obviously, it will be impor- uria, creatinine clearance, and duration of CTGF-N and TGF-␤1(r ϭ 0.671, P Յ tant to obtain follow-up data of these and diabetes was observed, which indicates 0.001) and weaker correlations between ϭ ϭ other microalbuminuric diabetic patients that in this analysis these three parameters CTGF and HbA1c (r 0.355, P 0.005) ␤ ϭ to investigate whether high plasma can be considered as independent factors. and between TGF- 1 and HbA1c (r CTGF-N levels might predict progression Nevertheless, a certain degree of interde- 0.477, P ϭ 0.029). The latter is in agree- of microalbuminuria to DN. pendence might be assumed because DN ment with previous observations (5) in a The major finding in forward step- tends to develop about 10–15 years after much larger population of diabetic sub- wise regression analysis was that plasma the onset of type 1 diabetes, starting with jects. Forward stepwise regression analy- CTGF-N levels were associated with albu- microalbuminuria, followed by protein- sis, however, revealed no significant minuria, creatinine clearance, and dura- uria and later progressive loss of renal addition of HbA1c to the prediction of tion of diabetes. The strongest correlation function. plasma CTGF-N levels from albuminuria, was found with albuminuria, one of the Although CTGF has originally been creatinine clearance, and duration of dia-

1168 DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004 Roestenberg and Associates betes. This suggests that, as is known for 11. Riser BL, Denichilo M, Cortes P, Baker C, Acknowledgments— This work was sup- the development of diabetes complica- Grondin JM, Yee J, Narins RG: Regulation ported by the Dutch Diabetes Fund (project of connective tissue growth factor activity tions, CTGF expression might be deter- grant 2000.00.058) and the Dutch Kidney mined largely by factors other than in cultured rat mesangial cells and its ex- Foundation (project grant NSN PC91). pression in experimental diabetic glomer- glycemic control alone. Dr. E. de Koning (Department of Vascular ulosclerosis. J Am Soc Nephrol 11:25–38, Given its biological activity in terms Medicine, UMC Utrecht) is gratefully ac- 2000 of fibrosis and matrix accumulation, it is knowledged for critical reading and helpful 12. Wahab NA, Yevdokimova N, Weston BS, possible that CTGF is not only a marker, discussion. Roberts T, Li XJ, Brinkman H, Mason RM: but also a pathogenic factor in the devel- Role of connective tissue growth factor in opment of DN and other complications. the pathogenesis of diabetic nephropathy. CTGF is involved in the induction of the References Biochem J 359:77–87, 2001 expression of ECM components by high- 1. Border WA, Noble NA: Interactions of 13. Gilbert RE, Akdeniz A, Weitz S, Usinger transforming growth factor-␤ and angio- WR, Molineaux C, Jones SE, Langham glucose concentrations in renal mesangial tensin II in renal fibrosis. Hypertension 31: RG, Jerums G: Urinary connective tissue cells and fibroblasts in vitro (12,26). Un- ␤ 181–188, 1998 growth factor excretion in patients with der diabetic conditions, TGF- and IGF-1 2. Sheetz MJ, King GL: Molecular under- type 1 diabetes and nephropathy. Diabe- are upregulated, while bone morphoge- standing of hyperglycemia’s adverse ef- tes Care 26:2632–2636, 2003 netic protein (BMP) is downregulated fects for diabetic complications. JAMA 14. Twigg SM, Chen MM, Joly AH, Chakrapani (3,27). Both inhibition of TGF-␤ and sup- 288:2579–2588, 2002 SD, Tsubaki J, Kim HS, Oh Y, Rosenfeld plementation of BMP-7 protect against 3. Flyvbjerg A: Putative pathophysiological RG: Advanced glycosylation end products (progression of) experimental DN role of growth factors and cytokines in up-regulate connective tissue growth factor (28,29). CTGF can bind TGF-␤, BMPs, experimental diabetic kidney disease. Dia- (insulin-like growth factor-binding protein- betologia 43:1205–1223, 2000 related protein 2) in human fibroblasts: a and IGF-1 and is also able to influence 4. Gupta S, Clarkson MR, Duggan J, Brady potential mechanism for expansion of ex- their activity. It enhances IGF-1–induced HR: Connective tissue growth factor: tracellular matrix in diabetes mellitus. Endo- secretion of collagens I and III, and it in- potential role in glomerulosclerosis and crinology 142:1760–1769, 2001 creases the profibrotic activity of TGF-␤ tubulointerstitial fibrosis. Kidney Int 58: 15. Blom IE, van Dijk AJ, Wieten L, Duran K, while inhibiting BMP activity (26,30,31). 1389–1399, 2003 Ito Y, Kleij L, deNichilo M, Rabelink TJ, Thus, CTGF, induced by hyperglycemia, 5. Chaturvedi N, Schalkwijk CG, Abraha- Weening JJ, Aten J, Goldschmeding R: In AGEs, or other pathways, might have an mian H, Fuller JH, Stehouwer CD: Circu- vitro evidence for differential involve- important impact on tissue remodeling lating and urinary transforming growth ment of CTGF, TGFbeta, and PDGF-BB in factor ␤1, Amadori albumin, and compli- mesangial response to injury. Nephrol Dial and fibrosis by its contribution to the di- cations of type 1 diabetes: the EURODIAB Transplant 16:1139–1148, 2001 abetes-related growth factor imbalance. prospective complications study. Diabetes 16. Hashimoto G, Inoki I, Fujii Y, Aoki T, Given the impressive elevation of urinary Care 25:2320–2327, 2002 Ikeda E, Okada Y: Matrix metalloprotein- CTGF excretion in DN, it is possible that 6. Pfeiffer A, Middelberg-Bisping K, Drewes ases cleave connective tissue growth fac- the nephropathy-associated increase of C, Schatz H: Elevated plasma levels of tor and reactivate angiogenic activity of plasma CTGF reflects, at least in part, in- transforming growth factor-␤1 in NIDDM. vascular endothelial growth factor 165. creased renal production in addition to Diabetes Care 19:1113–1117, 1996 J Biol Chem 277:36288–36295, 2002 the expected effect of reduced renal clear- 7. Bradham DM, Igarashi A, Potter RL, Gro- 17. Ball DK, Surveyor GA, Diehl JR, Steffen ance (13). In this way, plasma CTGF tendorst GR: Connective tissue growth CL, Uzumcu M, Mirando MA, Brigstock factor: a cysteine-rich mitogen secreted by DR: Characterization of 16- to 20-kilo- might be involved in nephropathy- human vascular endothelial cells is re- dalton (kDa) connective tissue growth associated aggravation of systemic (and lated to the SRC-induced immediate early factors (CTGFs) and demonstration of particularly cardiovascular) complica- gene product CEF-10. J Cell Biol 114: proteolytic activity for 38-kDa CTGF in tions of diabetes. 1285–1294, 1991 pig uterine luminal flushings. Biol Reprod In summary, we have observed that in 8. Oemar BS, Werner A, Garnier JM, Do DD, 59:828–835, 1998 patients with type 1 diabetes, elevated Godoy N, Nauck M, Marz W, Rupp J, 18. Molitch ME, DeFronzo RA, Franz MJ, plasma CTGF-N levels were associated Pech M, Luscher TF: Human connective Keane WF, Mogensen CE, Parving HH: with DN and correlated with level of al- tissue growth factor is expressed in ad- Diabetic nephropathy (Position State- vanced atherosclerotic lesions. Circulation ment). Diabetes Care 26 (Suppl. 1):S94– buminuria, impairment of creatinine 95:831–839, 1997 S98, 2003 clearance, and poor glycemic control. 9. Murphy M, Godson C, Cannon S, Kato S, 19. Stehouwer CD, Fischer HR, van Kuijk Longitudinal studies will be needed to Mackenzie HS, Martin F, Brady HR: Sup- AW, Polak BC, Donker AJ: Endothelial further investigate the possible clinical pression subtractive hybridization identi- dysfunction precedes development of mi- value of CTGF detection as a marker for fies high glucose levels as a stimulus for croalbuminuria in IDDM. Diabetes 44: incipient or progressive diabetes compli- expression of connective tissue growth 561–564, 1995 cations and, in particular, DN. To assess factor and other genes in human mesan- 20. Tschoepe D, Ostermann H, Huebinger A, the possible importance of increased gial cells. J Biol Chem 274:5830–5834, Ziegler D, Wiefels K, Gries FA: Elevated 1999 platelet activation in type I diabetics with CTGF levels for development and pro- 10. Ito Y, Aten J, Bende RJ, Oemar BS, Ra- chronic complications under long-term gression of these disorders, animal studies belink TJ, Weening JJ, Goldschmeding R: near-normoglycemic control. Haemostasis will be required in which CTGF expres- Expression of connective tissue growth 20:93–98, 1990 sion and activity is manipulated against a factor in human renal fibrosis. Kidney Int 21. Kaplan KL, Owen J: Plasma levels of plate- background of diabetes. 53:853–861, 1998 let secretory proteins. Crit Rev Oncol He-

DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004 1169 Elevated plasma CTGF in diabetic nephropathy

matol 5:235–255, 1986 N, Ueki K, Aiello LP, King GL: Vascular Jim B, Hong SW, Isono M, Ziyadeh FN: 22. Riser BL, Cortes P, Denichilo M, Desh- endothelial growth factor induces expres- Reversibility of established diabetic glo- mukh PV, Chahal PS, Mohammed AK, sion of connective tissue growth factor merulopathy by anti-TGF-beta antibodies Yee J, Kahkonen D: Urinary CCN2 via KDR, Flt1, and phosphatidylinositol in db/db mice. Biochem Biophys Res Com- (CTGF) as a possible predictor of diabetic 3-kinase-akt-dependent pathways in ret- mun 300:16–22, 2003 nephropathy: preliminary report. Kidney inal vascular cells. J Biol Chem 275:40725– 29. Lund RJ, Davies MR, Hruska KA: Bone Int 64:451–458, 2003 40731, 2000 morphogenetic protein-7: an anti-fibrotic 23. Adler SG, Kang SW, Feld S, Cha DR, 26. Lam S, Verhagen NAM, van der Geest RN, morphogenetic protein with therapeutic Barba L, Striker L, Striker G, Riser BL, van Nieuwenhoven FA, Blom IE, Aten J, importance in renal disease. Curr Opin LaPage J, Nast CC: Glomerular mRNAs in Goldschmeding R, Daha MR, van Kooten Nephrol Hypertens 11:31–36, 2002 human type 1 diabetes: biochemical evi- C: Connective tissue growth factor and 30. Wang S, Denichilo M, Brubaker C, dence for microalbuminuria as a manifes- IGF-I are produced by human renal fibro- Hirschberg R: Connective tissue growth tation of diabetic nephropathy. Kidney Int blasts, and cooperate in the induction of factor in tubulointerstitial injury of dia- 60:2330–2336, 2001 collagen production by high glucose. Di- betic nephropathy. Kidney Int 60:96–105, 24. Caramori ML, Fioretto P, Mauer M: The abetes 52:2975–2983, 2003 2001 need for early predictors of diabetic ne- 27. Wang SN, Lapage J, Hirschberg R: Loss of 31. Abreu JG, Ketpura NI, Reversade B, De phropathy risk: is albumin excretion tubular bone morphogenetic protein-7 in Robertis EM: Connective-tissue growth rate sufficient? Diabetes 49:1399–1408, diabetic nephropathy. J Am Soc Nephrol factor (CTGF) modulates cell signalling 2000 12:2392–2399, 2001 by BMP and TGF-beta. Nat Cell Biol 4: 25. Suzuma K, Naruse K, Suzuma I, Takahara 28. Chen S, Carmen Iglesias-de la Cruz M, 599–604, 2002

1170 DIABETES CARE, VOLUME 27, NUMBER 5, MAY 2004