|Home | About | Journals | Submit | Contact Us | Français|
Objective. To study the membrane expression of endothelial protein C receptor (mEPCR) in the renal microvasculature in lupus nephritis (LN) as a potential marker of injury and/or prognostic indicator for response to therapy.
Methods. mEPCR expression was analysed by immunohistochemistry in normal kidney and in 59 biopsies from 49 patients with LN. Clinical parameters were assessed at baseline, 6 months and 1 year.
Results. mEPCR was expressed in the medulla, arterial endothelium and cortical peritubular capillaries (PTCs) in all biopsies with LN but not in the cortical PTCs of normal kidney. Positive mEPCR staining in >25% of the PTCs was observed in 16/59 biopsies and associated with poor response to therapy. Eleven (84.6%) of 13 patients with positive staining for mEPCR in >25% of the PTCs and follow-up at 6 months did not respond to therapy, compared with 8/28 (28.6%) with mEPCR staining in 25% PTCs, P = 0.0018. At 1 year, 10 (83.3%) of 12 patients with positive mEPCR staining in >25% of the PTCs did not respond to therapy (with two progressing to end-stage renal disease) compared with 8/24 (33.3%) with positive staining in 25% of the PTCs, P = 0.0116. Although tubulo-interstitial damage (TID) was always accompanied by mEPCR, this endothelial marker was extensively expressed in the absence of TID suggesting that poor response could not be attributed solely to increased TID. mEPCR expression was independent of International Society of Nephrology/Renal Pathology Society class, activity and chronicity indices.
Conclusion. Increased mEPCR expression in PTCs may represent a novel marker of poor response to therapy for LN.
The contribution of the vascular endothelium to the pathogenesis of renal injury has not been emphasized in lupus nephritis (LN). Despite potential biological insights and treatment strategies to be gained by studying the endothelium in LN, neither historic WHO classification, National Institutes of Health chronicity and activity indices (CI and AI, respectively) , nor recent International Society of Nephrology/Renal Pathology Society (ISN/RPS) 2003 pathological classifications of LN  specifically address the state of the microvasculature in their definitions. However, recent murine data based on microarray analysis suggest that endothelial activation is a feature observed in progressive glomerulosclerosis but not in non-progressive glomerulosclerosis .
The membrane endothelial protein C receptor (mEPCR), an integral membrane protein with both anti-inflammatory and anti-thrombotic properties, is expressed on endothelial cells and regulates the conversion of protein C to activated protein C by presenting it to the thrombin–thrombomodulin complex [4,5]. mEPCR is shed in a pathological state to a soluble form, sEPCR, increased levels of which have been reported in two lupus cohorts [6,7]. Patients with LN had significantly higher levels of sEPCR than those without nephritis .
mEPCR expression in kidney disease has been evaluated in models of sepsis and diabetes [8,9] but never addressed in LN. Accordingly, the current study was initiated to test the hypothesis that the presentation and course of LN is influenced by the renal microvasculature, as reflected by mEPCR expression. This was approached by immunohistological analysis of mEPCR expression in kidney biopsies in patients with all classes of LN, and by correlating the results of the immunostaining with response to therapy.
The study was approved by the Institutional Review Board of New York University (NYU) School of Medicine. The twelve subjects followed prospectively signed NYU IRB approved consent forms for the study. A total of 301 biopsies from 242 patients with a diagnosis of LN between January 2000 and April 2008 were identified. Patients were selected based on the following criteria. Inclusion criteria: (i) confirmation of a diagnosis of SLE as per the ACR ; (ii) confirmation of a diagnosis of LN based on renal biopsy; (iii) availability of renal biopsy tissue in the paraffin block for immunohistochemical analysis; (iv) at least 10 glomeruli in the renal biopsy; and (v) access to medical records. Exclusion criteria: inappropriate fixation of the renal tissue determined by the absence of staining of CD31 in PTCs or mEPCR in renal medulla and/or arterial endothelium. Fifty-nine formalin-fixed, paraffin-embedded renal biopsies from 49 patients were available. Ten patients had two biopsies.
Medical records were reviewed by two physicians blinded to the biopsy staining results. Evaluation of proteinuria was assessed by either 24 h urine protein/creatinine ratio (24 h proteinuria, if creatinine was unavailable) or spot protein/creatinine ratio. A complete responder was defined as a return to within 10% of the normal values of serum creatinine and proteinuria (<0.5 g per 24 h). A partial responder was defined as improving by 50% in previous endpoints without worsening (within 10% of any measurement) . Patients who did not meet either criterion were defined as non-responders. Data were available for assessment of clinical follow-up in 41 patients at 6 months and in 36 patients at 1 year. Two patients had follow-up data at 1 year with no data at 6 months while seven patients had follow-up at 6 months and no data at 1 year.
Slides, electron microscopy photographs and pathology reports were reviewed by a renal pathologist (L.B.) blinded to patients’ treatments and outcomes. Each case was classified using the ISN/RPS 2003 classification : AI and CI were calculated in proliferative forms . The extent of interstitial inflammation and chronic tubulo-interstitial damage (TID), including interstitial fibrosis and tubular atrophy, was calculated in all renal biopsies and recorded as a percentage of cortex involved by TID or inflammation per total amount of cortex. The amounts of TID and interstitial inflammation in the renal cortex were separately semiquantitatively scored as follows: 0 = none; 1 = involvement of 25%; 2 = involvement of 26–50%; 3 = involvement of 51–75%; 4 = involvement of 76%. IF results and electron microscopic findings were reviewed and the presence of deposits in the interstitum and/or tubular basement membrane was recorded and correlated with immunohistochemistry results.
Sections from renal biopsies and normal human kidney from surgical nephrectomies were stained with antibodies against EPCR (Mab, HEPCR 1489 provided by C.T.E.) (1 : 100) using an avidin–biotin method. Immunostaining for CD31 (Ventana Medical Systems, Tucson, AZ, USA) (1 : 100), an endothelial cell marker, was also performed as a positive control for antigenicity of endothelial cells and to evaluate density of peritubular capillaries (PTCs) in the renal cortex. The isotope for mEPCR (Mouse IgG1, SouthernBiotech, Birmingham, AL, USA) was used as negative control. Briefly, antigen retrieval was performed by using high pH Antigen Retrieval Kits (Dako, Glostrup, Denmark, Catalog No. S 3307). Intensity of immunostaining was scored from 0 to 3+, and staining was considered positive when 1+ and negative when 0 or trace (0.5+). The percentage of renal cortex with positive staining for mEPCR in PTCs per total amount of cortex was then calculated. In addition, the extent of renal cortex with positive staining for mEPCR in PTCs was semiquantitatively scored as follows: 0 = no positive staining; 1 = involvement of 25% of the cortical PTCs; 2 = involvement of 26–50% of the cortical PTCs; 3 = involvement of 51–75% of the cortical PTCs; 4 = involvement of 76% of the cortical PTCs . Intra-rater variability was assessed by re-evaluation of the immunostaining by the same renal pathologist (L.B.), blinded to the clinical outcome and to the previous reading of the immunohistochemistry, 2 months following the initial analysis. In only two of the 59 biopsies was there a difference in the mEPCR score; and in both, the second score was used. Inter-rater variability of mEPCR score was addressed by having a second renal pathologist (D.B.T.) independently and under blinded conditions read and evaluate mEPCR immunostaining in the identical biopsies. In only four of 59 biopsies were there differences in the mEPCR score. All discrepancies were discussed and a consensus reached in the absence of any clinical data.
Given the moderate sample size of this study, partial and complete responders were combined and analysed together as responders. Justification for this approach is supported by the finding that a partial remission in LN is generally associated with a significantly better patient and renal survival compared with no remission . Only patients with Classes III, IV, V and mixed membranoproliferative LN were included in the responder analysis. Furthermore, for the 10 patients with two biopsies, only data at the time of the second biopsy were considered. Fisher's exact test was performed to evaluate bivariate associations between categorical variables. Student's t-test or the Mann–Whitney test, depending on the distribution of the data, was used for between-group comparisons of continuous variables. Correlations were estimated with the Spearman rank correlation coefficient. Bivariate analyses were not adjusted from multiple comparisons because these results were considered to be descriptive. Multivariable logistic regression models were also fitted to the data to adjust for potential confounders in assessing the association between mEPCR and responder status. A P-value <0.05 was considered to be significant.
Demographic characteristics of the 49 patients are shown in Table 1. The mean ± s.d. of age at the time of biopsy was 34.5 ± 10 years. Females comprised 88% of the patients. Minorities comprised 76% of the population. The mean ± s.d. creatinine and glomerular filtration rate (GFR) at the time of biopsy was 1.1 ± 0.6 and 83.1 ± 37.8, respectively. Thirty-seven percent of the patients had a GFR<60 ml/min. Low complement levels and elevated anti-dsDNA antibodies were seen in 75% and 77% of the patients, respectively. aPLs, defined as anti-cardilolipin IgG or IgM >40 or presence of LAC, were present in 18% of the patients. Thirty-five per cent had hypertension at the time of biopsy and 47% had nephrotic range proteinuria. In the majority of patients, induction therapy consisted of steroids plus either cyclophosphamide or mycophenolate mofetil. Of the 41 patients with clinical follow-up at 6 months, 19 were non-responders and 22 were responders (13 complete; 9 partial). Of the 36 patients in whom 1 year follow-up was available, 18 were non-responders and 18 responders (15 complete responders, 3 partial).
The 59 biopsies from 49 patients encompassed all ISN/RPS classes, the majority of which were proliferative forms (44/59 biopsies). In proliferative LN, the mean ± s.d. AI was 6.97 ± 4.00 (range 1–15); the mean ± s.d. CI was 2.11 ± 2.10 (range 0–8). Twenty-four of the 49 biopsies revealed deposits detected by IF or ultrastructural analysis in the tubular basement membranes and/or in the interstitium. The mean ± s.d. percentage of cortex involved by TID was 13.0 ± 14.7%, which corresponds to an average TID score of 1.0 ± 0.7. The mean ± s.d. percentage of cortex involved with inflammation was 12.4 ± 12.2%, which corresponded to an average inflammation score of 0.8 ± 0.6. Microthrombi were observed in only five biopsies.
There was no positive staining for mEPCR in cortical PTCs in the normal adult kidneys used as control (Fig. 1). In contrast, three biopsies with LN stained with mEPCR had a score of 0, 31 had a score of 1, five had a score of 2, six had a score of 3 and four had a score of 4. Although mEPCR staining was observed in PTCs in all areas where TID was present, TID involved 25% of the cortex in 86% of all biopsies, suggesting that the extent of mEPCR staining was not simply a reflection of TID. Furthermore, equivalent CD31 staining in serial biopsy sections supported that the extent of mEPCR expression was not due to differences in the representation of blood vessels (Fig. 1).
Staining for mEPCR in the glomeruli did not adhere to any definable pattern; most of the glomeruli were negative, whereas others revealed positive staining in the areas where large immune complex deposition were detected, indicating a probable non-specific binding of the secondary antibody. Positive staining for mEPCR in the PTCs was seen in the medulla, when present, of normal adult kidney and in LN biopsies. High background in the tubular cell cytoplasm was often observed in cases with severe proteinuria, representing non-specific binding of the secondary antibody to protein reabsorption droplets. The isotype showed mild staining in tubular cells, considered non-specific background, but was consistently negative in endothelial cells of PTCs of normal adult kidney and in all biopsies (Fig. 1).
mEPCR staining in cortical PTCs was seen in all classes of LN. There was no correlation between mEPCR score and AI (r = 0.03, P = 0.87) and CI (r = 0.20, P = 0.25). There was also no significant difference between mEPCR score in patients with immune complexes in the tubulo-interstitium (1.58) compared to those without (1.48), P = 0.87. There was a trend between mEPCR score in patients with microthrombi on the renal biopsy (2.4) and those without (1.4), P = 0.058. There was a statistical correlation between mEPCR score and TID score (r = 0.48, P = 0.0005). This was expected because mEPCR expression was always present in areas of TID. However, as noted above, mEPCR expression was also independent of TID since intense mEPCR staining was clearly observed in normal appearing cortical PTCs and thus the extent of mEPCR was not solely accounted for by TID (Fig. 2). In limiting the correlation analysis of mEPCR and TID to those patients with an mEPCR score 2, there was no statistically significant correlation (r = 0.17, P = 0.71). There was a correlation between mEPCR expression and interstitial inflammation (r = 0.35, P = 0.0125).
The baseline clinical and pathological variables of the 41 patients with 6 months of follow-up are shown in Tables 2 and and3.3. Of these, 13 (31.7%) had an mEPCR score 2 and 28 (68.3%) had a score of <2. At 6-month follow-up, 19/41 (46.3%) patients were non-responders. Eleven of 13 (84.6%) patients with an EPCR score of 2 were non-responders vs 8/28 (28.6%) patients with an EPCR score of <2, P = 0.0018 (Fig. 3). This association remained highly significant even after excluding the six patients with a TID score of 2. Specifically, 8/10 (80%) of biopsies with an mEPCR score of 2 were non-responders compared with 7/25 (28%) of biopsies with an mEPCR score of <2, P = 0.0082.
The durability of the association between poor response and mEPCR was supported by the observation that patients with an mEPCR score of 2 on biopsy continued to be classified as non-responsive at 1 year (Table 3). Specifically, 10/12 (83.3%) patients with an mEPCR score of 2 were non-responders at 1 year compared with only 8/24 (33.3%) with an mEPCR score of <2 (P = 0.0116). This association remained when patients with high TID scores were excluded (Fig. 3). Overall, two non-responders with an mEPCR score of 2 progressed to end-stage renal disease within 1 year compared with none of those with an mEPCR score of <2.
The number of patients with high levels of staining is limited. However, of the nine patients with an mEPCR score of 3, seven were non-responders at both 6 months and 1 year. All four patients with a score of 4 were non-responders at 6 months and three remained unresponsive at 1 year.
The only demographic criteria associated with an absence of response at both 6 and 12 months was Hispanic race. There was no difference in any laboratory parameter although as expected there was a trend in the percentage of patients with a GFR <60 in the non-responder group at 6 and 12 months. Patients with hypertension were more likely to be non-responders at 1 year although there was no difference in the use of angiotensin-converting enzyme inhibitors or angiotensin receptor blockers between non-responders and responders (data not shown). There was no difference in AI, CI, presence of microthrombi or inflammation scores between responders and non-responders. Although there was a higher TID score in the non-responders compared with responders, a TID score of 2 did not distinguish between groups.
Multivariable logistic regression models were also fitted to the data to evaluate whether the association between mEPCR score and responder status retained significance after adjustment for potential confounders. All variables which were significant in bivariate analyses at the 0.20 level were initially considered for inclusion as covariates in the model, and those which were significant at P < 0.05 and/or modified the estimated coefficient for mEPCR score by >15% in either direction were retained in the final model. Results indicate that mEPCR remains significantly associated with responder status at 6 months after adjusting for race, ISN/RPS Class, GFR and TID score [adjusted odds ratio (ORadj) = 3.5, P = 0.039]. The OR can be interpreted as the increase in the odds of response to therapy for each unit decrease in mEPCR score. With the 12-month follow-up data, the magnitude of the covariate adjusted association between mEPCR and responder status was nearly identical to the 6-month estimates, but the results were no longer significant because of the smaller sample size (ORadj = 3.6, P = 0.15).
Ten patients had serial biopsies due to clinical progression or disease relapse. Eight patients had data available 6 months after the first biopsy (all ten had follow-up after the second). Six of eight patients were responders (one was Class II) and had an mEPCR score of <2. Of the two non-responders one had an mEPCR score of 2. The mEPCR score increased in the second biopsy in six patients, remained the same in three and decreased in one.
An mEPCR score of 2 predicted response in 6/8 patients with 6 months follow-up after both biopsies. In one patient with Class III LN, the initial biopsy had an mEPCR score of 1 followed by complete response at 6 months. A repeat biopsy 2.5 years later revealed LN Classes IV and V and an mEPCR score of 2, followed by no response to therapy at 6 months.
In this study, immunohistological assessment revealed mEPCR expression in the cortical PTCs of biopsies with LN and not in normal human kidneys. mEPCR expression in the PTCs was independent of biopsy class, AI and CI. Positive staining in >25% of the PTCs in the renal cortex distinguished patients who responded to therapy at both 6 and 12 months from non-responders. Although TID was always accompanied by mEPCR, this novel endothelial marker was extensively expressed even in the absence of TID suggesting that a poor response could not be attributed solely to increased TID. End-stage renal disease was rare and occurred only in patients whose biopsies had an mEPCR score of 2.
The relationship between mEPCR expression and renal disease progression is intriguing since mEPCR is generally considered a protective molecule based on its role in both inflammation and coagulation. mEPCR binds protein C, presenting it to the thrombin–thrombomodulin complex, thus regulating its conversion to activated protein C . Conditions that reduce surface expression of mEPCR on endothelial cells attenuate the efficiency of protein C activation . The decrease in mEPCR is due to a metalloproteinase-dependent cleavage which splits the molecule into a soluble form, sEPCR .
Given the prediction that shed mEPCR impairs the integrity of the endothelium and places the net balance of this protective protein in biological ‘arrears’, a decrease in mEPCR expression was the predicted result in patients with progressive renal injury. Thus, finding increased mEPCR expression in the endothelium of PTCs, compared with controls, was unexpected.
The seemingly paradoxical finding of increased mEPCR in LN is provocative but not without precedent. It has been reported that mEPCR expression in the glomerular and interstitial microvasculature was increased in rodents with experimentally induced sepsis . In this model, sepsis was associated with a depletion of activated protein C, the consequence of which may be a compensatory increase in mEPCR as demonstrated in the diseased renal parenchyma. Similarly, increased mEPCR was also seen in renal biopsies from patients with acute kidney injury . One explanation for the findings herein is that increased mEPCR may represent a thwarted attempt at endothelial defence. An alternative hypothesis is that in LN circulating and/or deposited immune complexes activate the classical complement pathway generating C4b binding protein, which in turn complexes with protein S impairing its ability to generate activated protein C . Against this explanation is the absence of a correlation between mEPCR expression and local deposits of immune complexes or complement. However, it remains possible that circulating immune complexes, mimicking the effects of sepsis, may activate the endothelium of the microvasculature. Finally, as in sepsis, LN may induce a low protein C state.
Biomarker studies have corroborated the involvement of the endothelium in murine and human LN. In a murine study, severe LN was associated with high vascular cell adhesion molecule 1 levels in urine as well as increased expression in the tubules and vascular endothelium in LN . In humans, urine levels of adiponectin were significantly elevated prior to and during renal flares . In addition, adiponectin was expressed on endothelial surfaces in the renal microvasculature in patients with LN . Urine levels of the monocyte chemoattractant protein-1, produced by endothelial cells, have been shown to increase significantly during renal flares with decreases parallelling response [19,20]. Increased nitric oxide production, generated in part by vascular endothelial cells, has been associated with renal damage and poor response to therapy .
Several shortcomings are acknowledged with regard to interpretation of the overall data. This study was largely retrospective and the number of patients with biopsy tissue and data for clinical evaluation was limited. In addition, compliance with oral medication was difficult to address and patient numbers were too small to gain insight into the potential benefit of a specific therapy. The use of various treatments, each with a different response rate and time of response, further dictates caution in interpreting these findings.
In summary, positive mEPCR staining >25% in cortical PTCs is associated with a poor renal response to standard therapy. While these results suggest a contribution of the endothelium of the renal parenchyma to the pathophysiology of more progressive LN, further studies are needed to distinguish whether the endothelium plays an ‘active’ or ‘reactive’ role. Larger prospective studies are needed to affirm the significance of mEPCR expression as a novel biomarker of unanticipated renal progression and to address the utility of longitudinally measuring sEPCR in the serum and urine.
The authors would like to thank Amy Lawless for help in preparing this manuscript.
Funding: This work was funded by an SLE Foundation, Inc. grant to P.M.I., NIH/NIAMS grant R01 AR055088-01 to R.M.C. and an Alliance for Lupus Research pilot grant to R.M.C.
Disclosure statement: The authors have declared no conflicts of interest.