UvA-DARE (Digital Academic Repository) Determinants of acute and chronic renal allograft injury Kers, J. Link to publication

Transcription

1 UvA-DARE (Digital Academic Repository) Determinants of acute and chronic renal allograft injury Kers, J. Link to publication Citation for published version (APA): Kers, J. (2016). Determinants of acute and chronic renal allograft injury General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons). Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible. UvA-DARE is a service provided by the library of the University of Amsterdam ( Download date: 20 Sep 2018

2 Chapter 13 Intragraft tubular Vimentin and CD44 expression correlate with long-term renal allograft function and interstitial fibrosis and tubular atrophy Jesper Kers 1 Yi-Chun Xu-Dubois 2,3 Eric Rondeau 2,3 Nike Claessen 1 Mirza M. Idu 4 Joris J.T.H. Roelofs 1 Fréderike J. Bemelman 5 Ineke J.M. ten Berge 5 Sandrine Florquin 1 1 Department of Pathology, Academic Medical Center, Amsterdam (NL) 2 Institut National de la Santé et de la Recherche Médicale (INSERM) U702, Hôpital Tenon, Assistance Publique - Hôpitaux de Paris, Paris (FR) 3 Department of Nephrology and Kidney Transplantation, Hôpital Tenon, Assistance Publique - Hôpitaux de Paris, Paris (FR) 4 Department of Surgery, Academic Medical Center, Amsterdam (NL) 5 Department of Internal Medicine, Renal Transplant Unit, Academic Medical Center, Amsterdam (NL) Transplantation, 2010, PMID:

3 CHAPTER 13 Abstract Development of interstitial fibrosis and tubular atrophy (IF/TA) is the main histologic feature involved in renal allograft deterioration. The aim of this study was to validate whether de novo tubular expression of CD44 (transmembrane glycoprotein) and vimentin (mesenchymal cell marker), both involved in renal fibrosis, can operate as surrogate markers for late IF/TA and renal function. Furthermore, we wanted to establish the interrater reproducibility for the scoring system, which can be a problem in histologic assessments. Six-month protocol renal allograft biopsies (n=30 for matching 12 months estimated glomerular filtration rate (egfr) from which 20 matched the 12-month protocol biopsy) were immunostained for CD44 and vimentin, semiquantitatively scored by three observers of two centers, and correlated with IF/TA and egfr at 12 months. The interobserver agreement was good for CD44 (Kendall s W-coefficient: 0.69; P<0.001) and vimentin (Kendall s W-coefficient: 0.79; P<0.001). CD44 and vimentin expression at 6 months were significantly correlated with IF/TA (r=0.481 for CD44 and r=0.619 for vimentin) and egfr (r=0.569 for CD44 and r=0.376 for vimentin) at12months. Summarizing, de novo tubular expression of CD44 and vimentin can function as surrogate marker for IF/TA and egfr at 12 months. Further area under receiver operator characteristic curve analysis has to establish the predictive value for both biomarkers. 252

4 Tubular vimentin and CD44 predict chronic graft dysfunction Introduction Since the introduction of potent immunosuppressive therapies, the incidence of acute allograft rejection has decreased, and accordingly, first year graft survival rates have increased up to 90% to 95%. Long-term graft survival has also increased but to a lesser extent (Meier-Kriesche et al., 2004). Ten years after transplantation, approximately 40% of all renal grafts failed, mainly as a consequence of the development of chronic allograft nephropathy (CAN) (Paul, 1999; Schweitzer et al., 1991). CAN is not a classifying diagnosis but a nonspecific term used to describe chronic graft scarring characterized by interstitial fibrosis and tubular atrophy (IF/ TA) that may result from both immune- and nonimmune-mediated injuries to the graft (Colvin and Thomson, 2002; Solez et al., 1998). Renal chronic allograft damage may evolve after transplantation without being predicted by deteriorating renal function (Hertig et al., 2008; Kaplan et al., 2003; Yilmaz et al., 2007). Therefore, there is a need to find reliable biomarkers that predict the long-term clinical graft outcome at an early stage, when intervention may still be beneficial. The accumulation of myofibroblasts in the interstitium is a key event in the development of fibrosis. Myofibroblasts may originate from the bone marrow-derived cells, stromal cells, fibrocytes, endothelium, activation of local fibroblasts, and tubular epithelial cells (TECs) by a phenomenon called epithelial-to-mesenchymal transition (EMT) (Iwano et al., 2002). Recently, a review series was published, discussing the role for epithelial cell plasticity in various disease models. The consensus outlined in this review series was the existence of three distinct pathways of EMT, named type 1 (primitive epithelial cells that differentiate into mesenchymal cells, which in turn differentiate into secondary epithelium), type 2 (secondary epithelia that differentiate into fibroblasts), and type 3 (epithelial tumor cells that become metastatic tumor cells and eventually form secondary tumor nodules) (Acloque et al., 2009; Kalluri, 2009; Kalluri and Weinberg, 2009; Zeisberg and Neilson, 2009). 13 The presence of EMT-related markers was previously investigated in renal allograft biopsies. A relationship has been shown between the expression of the type II EMT markers S100A4 and a-smooth muscle actin (SMA) and serum creatinine in transplant biopsies exhibiting IF/TA (Vongwiwatana et al., 2005). Hertig et al. (Hertig et al., 2008) found that expression of vimentin by TECs on protocol renal biopsies 253

5 CHAPTER 13 correlated with the progression of the Banff s IF/TA score. At present, it is still a matter of debate whether activated mesenchymal cells are able to break-down the tubular basement membrane and transmigrate into the interstitium or whether the mesenchymal cells contribute to fibrosis without actual transmembrane migration. Therefore, Hertig et al. regarded the dedifferentiation of TECs as an activational state of the epithelium for fibrogenesis, without drawing any conclusion on their role in myofibroblast generation or differentiation. One of the molecules that may be instrumental for EMT and fibrosis is CD44. CD44 comprises a large family of surface expressed type I transmembrane glycoproteins involved in multiple and complex biologic processes including inflammation, tumorigenesis, migration, and fibrosis (Heldin et al., 2008; Jothy, 2003; Marhaba and Zöller, 2004; Martin et al., 2003; Naor et al., 2005; Puré and Cuff, 2001; Siegelman et al., 1999). Healthy kidneys do not express CD44 on their tubules (Roy-Chaudhury et al., 1996). Ischemia-reperfusion injury is a major factor for postsurgical delayed graft function (DGF) and the development of CAN (Gueler et al., 2004; Shoskes and Cecka, 1998). Our previous experimental studies established a crucial role for CD44 in the cascade of inflammation, tubular damage, and fibrosis, which are hallmarks of deterioration of graft function. Disruption of CD44 led to decreased neutrophil influx in the postischemic mouse kidney and preserved renal function (Rouschop et al., 2005). CD44 deficiency in the mouse unilateral ureteral obstruction model increased the tubular damage but decreased the amount of IF (Rouschop et al., 2004). Persistent tubular damage and progression toward IF after unilateral ureteral obstruc- tion have been related to loss of peritubular capillaries (Bohle et al., 1996). Disruption of CD44 was associated with a relative preservation of the capillary network (Rouschop et al., 2006a). In IgA nephropathy, CD44 expression in the tubulointerstitial compartment correlated with the degree of IF/TA (Florquin et al., 2002; Qiaoling et al., 2009). Also in other types of primary glomerular injury, tubular expression of CD44 correlated with renal function and outcome (Daniel et al., 2001; Kirimca et al., 2001; Nakamura et al., 2005). In renal transplantation, CD44 was up-regulated on tubules during acute rejection episodes, and its expression was related to the severity of rejection (Rouschop et al., 2006b; Sarioğlu et al., 2004). Altogether, these studies implicate the involvement of CD44 in acute graft injury and a co-operative role in the progression toward renal scarring. 254

6 Tubular vimentin and CD44 predict chronic graft dysfunction The aim of this study is 3-fold: (1) to test the reproducibility of the scoring for vimentin and CD44 between independent observers of different centers; (2) to validate early vimentin expression on tubules as a surrogate marker for long-term function and scarring in an independent set of protocol renal allograft biopsies; and (3) to assess the value of CD44 expression on tubules as a biomarker for progression to IF/TA. Materials & Methods Patient population and biopsies In total, 126 renal transplant patients (older than 18 years) were enrolled in a prospective study conducted by the Academic Medical Center in which they received tailored calcineurin inhibitor regimens as described previously (Rowshani et al., 2006). Patients received antirejection treatment of cyclosporine- or tacrolimusbased immunosuppression regimens. Protocol renal biopsies were obtained at 6 and 12 months after transplantation. After processing of the diagnostics department, only 30 patients had 6 months biopsy fulfilling the adequacy criteria of Banff working classification 2005 (seven glomeruli and one artery per section), and 20 of them had matched 6 and 12 months adequate protocol biopsies. These 30 patients were included in this study. Of these 30 patients, 14 had matched implantation and 6 months biopsies and 11 had matched implantation and 12 months biopsies. All biopsies were scored according to the Banff 05 working classification update concerning CAN (Solez et al., 2007). Immunohistochemistry Monoclonal mouse antibodies against a-sma (1:800 diluted monoclonal antibody [mab] 1A4, Dako, Glostrup, Denmark), vimentin (1:400 diluted mab V9, Zymed Laboratories Inc., South San Francisco, CA), and the standard form of CD44 (1:2500 diluted mab Hermes-3, kindly provided by Dr. S. Pals, Academic Medical Center, Amsterdam, The Netherlands) were used to visualize the expression patterns in formalin-fixed, paraffin-embedded implantation, 6 and 12 months protocol biopsies. The sections were deparaffinized and rehydrated followed by peroxydase activity block. Vimentin and a-sma sections were pretreated with a ph 6.0 citrate buffer at 98 C for 10 min. Sections were incubated with aforementioned antibodies for 1 hr at room temperature, followed by incubation with secondary anti-mouse anti- bodies

7 CHAPTER 13 for 30 min, after which peroxydase activity was detected by 3,3-diaminobenzidine (Dako). All renal biopsies were stained at the pathology department of the Academical Medical Center, Amsterdam and afterward transported to Tenon Hospital Paris to assess interobserver agreement. Scoring of immunohistochemical stainings Three observers (J.K., S.F., and Y.-C.X.) of two independent medical centers scored all 6 and 12 months sections for CD44 and vimentin. TECs expressing basolateral CD44 or vimentin were considered positive. The tubules were scored in a blind fashion as described previously (Hertig et al., 2006; Rouschop et al., 2006b): total cortex surface was analyzed, except for the subcapsular region. Score 0: none; score 1: less than 10%; score 2: 10% to 24%; score 3: 25% to 50%; and score 4: more than 50%. The average score of the observers decided the final score. Staining patterns of CD44 and vimentin that are representative of semiquantitative scores are shown in Figure 1. The interobserver agreement was calculated on the total set of 6 and 12 months biopsies (n=50). Interstitial expression of a-sma with Figure 1 Representative immunohistochemical staining patterns for CD44 and vimentin on protocol renal transplant biopsies. (A) No expression of vimentin by the tubules (score 0). (B) Ten percent to twenty-five percent (score 2) of tubules show expression of vimentin. (C) More than 50% of tubules express vimentin (score 4). (D) No expression of CD44 by the tubular epithelial cells (score 0). Cells in the interstitial compartment express CD44. (E) Basolateral expression of CD44 in 10% to 25% of tubules (score 2). (F) More than 50% of tubules express CD44 (score 4). Magnification X20, inserts X

8 Tubular vimentin and CD44 predict chronic graft dysfunction exclusion of vessels and the subcortical area at 6 months biopsies was quantified with the use of a Leica DM5000 microscope (Leica Microsystems B.V., Rijswijk, The Netherlands) with a Nuance multispectral imaging camera (CRI, Woburn, MA) and reported as percentage positive a-sma expression of the total biopsy. Definition of chronic graft dysfunction according to the banff score Originally described in the 97 Banff report are the histologic features that define chronic allograft dysfunction: mesangial matrix increase (mm), vascular fibrous intimal thickening (cv), transplant glomerulopathy (cg), arteriolar hyaline thickening (ah), TA (ct), and IF (ci) (Hertig et al., 2006). As reviewed in the Banff 05 meeting report, a combination of the IF/TA score defines the histologic score for chronic allograft dysfunction and replaces the misused term CAN (Solez et al., 2007). Definition of renal function Serum creatinine was measured at 6 and 12 months after transplantation, and the renal function was estimated using the modification of diet in renal disease estimated glomerular filtration rate (egfr) formula, discriminating for age, race, and gender. Statistical analysis On the total set of biopsies, the interobserver agreement was analyzed using the Kendall s W-coefficient of the Friedman s chi-square test with the use of STATA version 8 for Windows (Stata Corp, College Station, TX). The differences between implantation and 6 and 12 months clinical and histopathologic parameters were calculated using a Wilcoxon nonparametric signed ranks test. Correlation coefficients and significance were calculated with the use of Spearman nonparametric rank test and Mann-Whitney U rank test where appropriate. A two-tailed P less than 0.05 was considered as significance threshold. Statistical analysis was performed using SPSS statistics version 17.0 for Macintosh (SPSS Inc., Chicago, IL). Calculation of the necessary sample size for proper receiver operating characteristic (ROC) analysis was performed with the use of PASS statistical software (NCSS, Kaysville, UT)

9 CHAPTER 13 Results Patient characteristics In total, 30 patients were analyzed. Table 1 shows the demographic characteristics of the patients included in this study and the total patient cohort. Clinical and pathologic parameters The median score for IF/TA was 0 at 6 months and 1 at 12 months (P=0.021). The median value for serum creatinine was 134 mmol/ml at 6 months and 130 mmol/ml at 12 months (P=0.823). The median value for egfr was 49 ml/min at 6 months and 50 ml/min at 12 months (P=0.964). None of the biopsies presented with intimal arteritis. Thus, in the further analysis, this parameter was omitted. Table 1 Demographic and clinical characteristics of the patient cohort ESRD, end-stage renal disease. 258

10 Tubular vimentin and CD44 predict chronic graft dysfunction Interobserver agreement for CD44 and Vimentin score The interobserver agreement for CD44 showed a Kendall s W-coefficient of 0.69 (P<0.001). For vimentin, the interobserver agreement showed a Kendall s W-coefficient of 0.79 (P<0.001). The average score of the observers per biomarker defined the final score used to calculate the associa- tion with the clinical and histologic parameters. Tubular CD44 and Vimentin expression in correlation with Banff scores and graft function Normal kidneys do not express CD44 or vimentin on their tubules. However, at implantation, some tubules stained positive for CD44 and vimentin. CD44 tubular score did not significantly alter after transplantation (nonsignificant Wilcoxon signed rank test from implantation to 6 months and 6 to 12 months). Vimentin tubular score significantly increased during initial 6 months after transplantation but remained stable from 6 to 12 months (Z =2.79 with P<0.01 from implantation to 6 months). Neither CD44 nor vimentin tubular score at implantation correlated with IF/TA or egfr at 12 months after transplantation (data not shown). Table 2 summarizes the scores for tubular CD44 and vimentin expression at 6 months in correlation with acute and chronic Banff scores and renal function up to 12 months after transplantation. Tubular expression of CD44 at 6 months was correlated with tubulitis, IF/TA, and renal function (both serum creatinine and egfr) at 6 months. Tubular expression of vimentin at 6 months was correlated with interstitial inflammation, tubulitis, IF/ TA, and renal function at 6 months. At the 12 months protocol biopsy, CD44 and vimentin correlated with Banff s glomerulitis and IF/TA scores but not with interstitial inflammation, tubulitis, and renal function (data not shown). 13 To validate a possible predictive function for CD44 and vimentin, 6 months scores were related to the Banff scores and the renal function at 12 months. Both tubular CD44 and vimentin expression at 6 months correlated with IF/TA and renal function (serum creatinine and egfr) at 12 months. In addition, CD44 at 6 months was correlated with glomerulitis at 12 months. Figure 2 shows the association of tubular CD44 and vimentin score at 6 months with IF/TA score (2A) and egfr (2B) at 12 months. 259

11 CHAPTER 13 Table 2 CD44 and vimentin expression on renal tubules at 6 mo correlated to donor and recipient characteristics, perisurgical events, Banff scores, and renal function 1 Parameter s association is calculated with Spearman correlation test and shown as ρ-coefficient. 2 Parameter s association is calculated with Mann-Whitney U test and shown as Z-coefficient. * P<0.01, ** P<

12 Tubular vimentin and CD44 predict chronic graft dysfunction Figure 2 Raw data of the tubular CD44 (red squares and line) and vimentin score at 6 months (black squares and line) in correlation with the interstitial fibrosis and tubular atrophy (IF/TA) score at 12 months. (A) Tubular CD44 and vimentin expression at 6 months correlated to IF/TA score at 12 months after transplantation. (B) Tubular CD44 and vimentin expression correlated to the estimated glomeru- lar filtration rate (egfr) at 12 months after transplantation. Relationship between interstitial a-sma expression at 6 months and Banff scores, CD44, and Vimentin tubular scores and renal function Interstitial myofibroblasts have been previously associated with IF/TA (Vitalone et al., 2008). Interstitial a-sma expression at 6 months ranged from 8.19% to 59.63% with a median of 17.43% of the total cortical area and was correlated with interstitial inflammation (r=0.56 with P<0.01), CD44 tubular score (r=0.63 with P<0.01), and vimentin tubular score (r=0.54 with P<0.01) but not with IF/TA (r=0.29 with P=0.17) and egfr (r=-0.32 with P=0.12) at 6 months. Interstitial a-sma at 6 months was neither correlated with IF/TA (r=0.33 with P=0.11) nor correlated with the egfr (r=-0.17 with P=0.42) at 12 months. 13 Other parameters involved in renal transplant dysfunction Many surrogate markers able to predict transplant failure have been proposed in the past. We examined the correlation of conventional, most used surrogate markers with late IF/TA, and decline in renal function. Table 3 overviews the association of demographic and surgery specific data, early renal function, and Banff scores with IF/TA or egfr at 12 months. Renal function at discharge and 6 months after transplantation (both serum creatinine value and egfr), tubulitis, 261

13 CHAPTER 13 interstitial inflammation, and IF/TA scores were correlated with the extent of IF/TA at 12 months. Donor age, renal function (serum creatinine and GFR), glomerulitis, tubulitis, and IF/TA score at 6 months after transplantation were associated with a lower egfr at 12 months. Because of the relatively small patient cohort, it was not possible to per- form multivariable linear regression analysis to verify the relative prognostic value of each variable. Parameters that are associated with renal tubular CD44 and Vimentin expression at 6 months after transplantation Table 2 shows the association of CD44 and vimentin tubular score at 6 months with donor or recipient characteristics, clinical parameters, and Banff scores. Tubular CD44 expression at 6 months was correlated with renal function at any time point up to 6 months after transplantation and DGF (median 6 months CD44 score of Table 3 Correlation of early parameters with IF/TA and egfr at 12 mo after renal transplantation Parameters are displayed as Spearman s ρ-coefficient. * P < ** P <

14 Tubular vimentin and CD44 predict chronic graft dysfunction 1.42 and 2.33 non-dgf vs. DGF, respectively). None of the Banff parameters at implantation was associated with tubular CD44 expression at 6 months. Tubular vimentin expression at 6 months was significantly associated with DGF (median 6 months vimentin score of 1.67 and 3.42 non-dgf vs. DGF, respectively) and preceding egfr at any time point up to 6 months and also with the type of allograft donation (median 6 months vimentin score of 1.33 and 2.50 living vs. deceased donor, respectively). Also, no Banff parameter at implantation was associated with vimentin expression on the tubules at 6 months. Discussion The search for predictors of renal transplant dysfunction has been a quest for a long time. This makes the identification of biomarkers associated with long-term renal function and histologic decline necessary, at a time where intervention is still possible and before irreversible graft loss has taken place. Serum creatinine levels and egfr are insensitive measures that are generally not influenced during the early stage of IF/TA development. Indeed, it has been shown that two thirds of all renal transplants developed IF/TA without loss of renal function (Solez et al., 1998). In this article, we have validated the association of early de novo CD44 and vimentin expression on tubules with the graft outcome 12 months after renal transplantation. For both markers, there is an excellent reproducibility of scoring among three observers of two independent medical centers. In concordance with the results found by Hertig et al. (Hertig et al., 2008), tubular vimentin expression seems to be associated with long-term IF/TA and egfr. Newly introduced in this article is the association of tubular CD44 expression with the long-term decline in renal function and deterioration of graft histology. 13 The Banff 97 report on allograft nephropathy described a shift from diagnosis toward prediction of allograft outcome as a role for renal biopsies (Racusen et al., 1999). Major determinants for late allograft outcome are ischemia-reperfusion injury, DGF, and rejection (Kupin et al., 1997). At present, the exact mechanism behind graft deterioration is still unknown, but new insights in its etiology are obtained frequently. Because of the poor value of the renal function in predicting graft outcome (Yilmaz 263

15 CHAPTER 13 et al., 2007), the focus has shifted toward both histologic changes and immunologic regulation of tolerance and ongoing (subclinical) rejection. This has led to the search for surrogate markers that represent these particular fields of interest. Since 2005, the histologic endpoint of the allograft was defined as IF together with TA, which indicates an important role for TECs in the progression toward histologic allograft loss (Solez et al., 2007). As summarized in Table 4, many markers for functional or histologic endpoints have been proposed. Two thirds of patients that proceeded toward loosing their allografts showed chronic lesions on protocol biopsies as early as 3 months after transplantation (Solez et al., 1998), and therefore, biomarkers for egfr and IF/ TA might differ. Conventional markers such as Banff s chronic histologic changes (vascular intimal thickening, tubular injury, and interstitial fibrosis) were associated with late structural allograft loss. Notably, from the chronic parameters of the Banff working classification, in our study, only early IF/TA is associated with late IF/ TA. A major problem of histologic grading systems, including the Banff working classification, is the poor reproducibility among observers (Furness et al., 2003; Marcussen et al., 1995). In our study, both vimentin (Kendall s W=0.79) and CD44 (Kendall s W=0.69) scores are reproducible among three observers, which makes them applicable in clinical practice. Allograft loss is believed to be the result of a final common pathway that leads to atrophic nonfunctional tubules and apoptosis of TECs, together with interstitial scarring that irreversibly stiffens the graft (Solez et al., 2007). Recent studies focus mainly on etiologic features of IF/TA that are able to predict graft loss, where intervention might still be beneficial. This article highlights two surrogate markers that indicate an early activated state of the renal tubules [epithelial phenotypical changes as described by Hertig et al. (Hertig et al., 2008)] before becoming irreversibly damaged. Vongwiwatana et al. (Vongwiwatana et al., 2005) described de novo expression of the mesenchymal markers S100A4, fibroblast specific protein (FSP)-1, and vimentin on TECs in deteriorating and atrophic and in relatively intact tubules. This is also the case for vimentin and CD44 in this study (Figure 1), indicating a precursor role for epithelial phenotypical changes in tubular atrophy and damage. Accordingly, the profibrotic agent transforming growth factor (TGF)-β1 was associated with 264

16 Tubular vimentin and CD44 predict chronic graft dysfunction Table 4 Previously described markers for the prediction of renal transplant dysfunction 1 (Kupin et al., 1997; Nankivell et al., 2001, 2002; Nicholson et al., 1997; Pape et al., 2003; Sund et al., 2004) 2 (Nicholson et al., 1997; Nickerson et al., 1998; Pape et al., 2003; Sund et al., 2004) 13 graft dysfunction (Ishimura et al., 2001), which later proved to be a strong inducer of TEC dedifferentiation, as reviewed by Iwano and Neilson (Iwano and Neilson, 2004). Downstream of TGF-b, connective tissue growth factor, also seemed to be associated with interstitial fibrosis. ELISA analysis of the urine and quantitative realtime polymerase chain reaction for intragraft mrna of connective tissue growth factor was associated with IF/TA at the time of biopsy (Cheng et al., 2006). Interestingly, antivimentin antibodies are formed in non-human primates and in patients who have received an allograft (Jonker et al., 2005; Jurcevic et al., 2001). Jonker et al. showed that these antivimentin antibodies were formed at the time of rejection. It 265

17 CHAPTER 13 might be possible that the formation of antivimentin antibodies contributes to the development of TA and TEC apoptosis. The correlation between CD44 tubular expression on protocol biopsies and late renal allograft outcome is described in this article. CD44 is a complex glycoprotein involved in important physiologic and pathologic processes comprising cell-cell and cell-matrix interactions, wound healing or scarring, cell migration, lymphocyte extravasation and activation, and apoptosis or proliferation (Jones et al., 2000; Rouschop et al., 2004; Siegelman et al., 1999; van der Voort et al., 1999; Weiss et al., 2000). Healthy kidneys do not express CD44 or vimentin on their tubules, but we found that these markers are already present in some tubules of the implantation biopsy. Interestingly, cold ischemia time was not associated with the expression of tubular CD44 and vimentin at implantation (data not shown). The exact role of CD44 expression on TECs is complex and still only partly known. In vitro studies and experimental murine models of renal diseases showed that CD44 exerted protective effects on TECs but contributed to renal fibrogenesis at least in part through enhancement of hepatocyte growth factor and TGF-β1 signaling pathways (Rouschop et al., 2004). Heilman et al. (Heilman et al., 2010) showed that subclinical inflammation and subclinical rejection were independently associated with high IF/ TA scores at the 1-year protocol biopsy. Damaged epithelia are able to express danger signals or stress signals such as CD44 and vimentin that in turn elicit an inflammatory response to remove injured cells [reviewed in (Zitvogel et al., 2010)]. In line with these findings, we found that both CD44 and vimentin were correlated with inflammation at 6 months, which in turn was correlated with IF/TA and egfr at 12 months. The limitation of this study is the relatively small number of biopsies. To establish predictive values for tubular expression of CD44 and vimentin as surrogate markers for allograft dysfunction, ROC analysis must be performed. This would provide the opportunity to calculate a specific cutoff value for CD44 and for vimentin and distinct patients who are at high risk for bad allograft outcome. In addition, sensitivity and specificity of the test can be assessed to validate the predictive value of tubular expression of CD44 and vimentin (Tripepi et al., 2009). Unfortunately, because of the relatively small patient cohort in this study, ROC analysis will not be accurate 266

18 Tubular vimentin and CD44 predict chronic graft dysfunction and, thus, will have no additional value for this article. To perform proper ROC analysis, the minimum amount of patients in the IF/TA positive group (score 2) and the IF/TA negative group (score<2) must be 30 and 90, respectively. More research is needed to discover the implications of epithelial CD44 and vimentin expression on renal damage. This will provide insights in the cause of IF/TA lesion formation, and it might also identify new interventional possibilities. In summary, this article corroborates the association of tubular vimentin expression with allograft outcome as de- scribed previously by Hertig et al. (Hertig et al., 2008). For the first time, we introduce the association of CD44 expression on TECs with the long-term decline in renal function and deterioration of graft histology. The scoring of both markers is reproducible, which makes them applicable for clinical practice