The authors have the following interests. This study was partly funded by Baxter International, Inc. There are no patents, products in development or marketed products to declare. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors.
Conceived and designed the experiments: FRR SLA MKB IES ZKZ NB MDA. Performed the experiments: FRR ZKZ NB DB PF. Analyzed the data: FRR SLA MKB IES ZKZ NB DB PF SAK HRT. Contributed reagents/materials/analysis tools: SLA MKB IES ZKZ NB DB PF JL. Wrote the paper: FRR SLA MKB NB SS.
Encapsulating peritoneal sclerosis (EPS) is a devastating complication of peritoneal dialysis (PD), characterized by marked inflammation and severe fibrosis of the peritoneum, and associated with high morbidity and mortality. EPS can occur years after termination of PD and, in severe cases, leads to intestinal obstruction and ileus requiring surgical intervention. Despite ongoing research, the pathogenesis of EPS remains unclear. We performed a global transcriptome analysis of peritoneal tissue specimens from EPS patients, PD patients without EPS, and uremic patients without history of PD or EPS (Uremic). Unsupervised and supervised bioinformatics analysis revealed distinct transcriptional patterns that discriminated these three clinical groups. The analysis identified a signature of 219 genes expressed differentially in EPS as compared to PD and Uremic groups. Canonical pathway analysis of differentially expressed genes showed enrichment in several pathways, including antigen presentation, dendritic cell maturation, B cell development, chemokine signaling and humoral and cellular immunity (P value<0.05). Further interactive network analysis depicted effects of EPS-associated genes on networks linked to inflammation, immunological response, and cell proliferation. Gene expression changes were confirmed by qRT-PCR for a subset of the differentially expressed genes. EPS patient tissues exhibited elevated expression of genes encoding sulfatase1, thrombospondin 1, fibronectin 1 and alpha smooth muscle actin, among many others, while in EPS and PD tissues mRNAs encoding leptin and retinol-binding protein 4 were markedly down-regulated, compared to Uremic group patients. Immunolocalization of Collagen 1 alpha 1 revealed that Col1a1 protein was predominantly expressed in the submesothelial compact zone of EPS patient peritoneal samples, whereas PD patient peritoneal samples exhibited homogenous Col1a1 staining throughout the tissue samples. The results are compatible with the hypothesis that encapsulating peritoneal sclerosis is a distinct pathological process from the simple peritoneal fibrosis that accompanies all PD treatment.
Renal replacement therapy is currently restricted to renal transplantation, hemodialysis (HD), and peritoneal dialysis (PD). Peritoneal dialysis constitutes <10% of current renal replacement therapy in the US and in much of Europe, but up to 80% in Mexico and Taiwan
Encapsulating peritoneal sclerosis (EPS) is a rare but dangerous complication of peritoneal dialysis (PD). Mortality rates as high as 57%
The pathogenesis of EPS remains incompletely understood. Simple peritoneal fibrosis accompanies nearly all PD treatment, resulting in gradual impairment of ultrafiltration that can necessitate transition to HD
The absence of blood tests specific for EPS requires diagnosis based on clinical presentation, radiologic and histologic findings
Non-surgical therapeutic options for EPS are few, and randomized controlled trials non-existent
Models of peritoneal fibrosis in rats and mice
Therefore, we have performed a pilot study to compare transcriptomes of fresh-frozen peritoneal biopsy samples from EPS patients with those of PD patients without EPS, and with those of uremic patients (Uremic) prior to initiation of dialysis. We employed systems biology and interactive network analyses to identify pathways and interactive networks enriched with EPS-dysregulated genes, to establish the feasibility of using this approach to unravel pathophysiological mechanisms of EPS development in PD patients. This pilot study lays an empiric foundation for future investigations to understand biological mechanisms of EPS and to identify novel prognostic and therapeutic biomarkers.
All samples were obtained from the peritoneal biopsy registry at Robert-Bosch Hospital, Stuttgart, Germany. Human peritoneal tissue, blood and peritoneal dialysate was collected at time of surgery at Robert-Bosch Hospital. Written informed consent was obtained from Patients according to an approved protocol (#322/2009BO1, Ethik-Kommission, Eberhard-Karls-Universität Tϋbingen, Germany). The approved protocol included a patient information sheet explaining the purpose of the study and the envisioned use of the tissue specimen. All patients agreeing to participate in the study signed, in the presence of an authorized clinical investigator, a detailed written consent form previously approved by the “Ethik-Kommission” of Eberhard-Karls-Universität Tϋbingen, Germany. Patients were provided one day during which to reconsider their written consent. Surgeries included implantation or reimplantation of peritoneal dialysis (PD) catheters or, for EPS patients, enterolyses or peritonectomies. The diagnosis of EPS was based on clinical and radiological findings and on histological criteria that included fibrosis, fibroblast-like cells, exudation, cellularity and its variability, vessel density and its variability, acute or chronic inflammation, hemorrhage, mesothelial hyperplasia, fibrin deposits, presence of vasculopathy, mesothelial denudation, presence of acellular areas, presence of iron and/or calcium deposits, and osseous metaplasia
RNA was isolated from fresh-frozen tissues using Trizol reagent (Invitrogen, Carlsbad, CA). Frozen tissue samples were homogenized in 1 ml Trizol reagent using a rotor-stator homogenizer (Tissue Tearor, BioSpec Products, Bartlesville, OK) for 1 minute. After chloroform extraction of the homogenate, the aqueous RNA-containing supernatant was twice ethanol-precipitated and resuspended in nuclease-free water. RNA concentration was measured by UV spectrophotometry (NanoDrop ND-1000, NanoDrop Technologies, Wilmington, DE). RNA integrity was assessed using the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA).
Tissue RNA samples were DNAse-treated (DNA-free kit, Applied Biosystems, Carlsbad, CA, Foster City, CA), and 250 ng total RNA from each specimen was reverse-transcribed using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) for 10 min at 25°C, 120 min at 37°C, and 5 min at 85°C in a GeneAmp PCR System 2700 thermal cycler (Applied Biosystems). Resultant cDNA samples were diluted 10-fold.
qRT-PCR was performed in duplicate with TaqMan gene expression kits and TaqMan Fast Universal PCR master mix (2x) using the “fast protocol” performed on the 7500 Fast Real-Time PCR System (Applied Biosystems) in bar-coded MicroAmp 96-well plates (Applied Biosystems), per manufacturer's instructions. The gene products analyzed included Thrombospondin 1 (THBS1), Matrix Metalloproteinase 2 (MMP2), Leptin (LEP), Retinol-Binding Protein 4 (RBP4), Runt-Related Transcription Factor 2 (RUNX2), Intercellular Adhesion Molecule 1 (ICAM-1), α Smooth Muscle Actin (ACTA2), Fibronectin 1 (FN1), Collagen 1α1 (Col1a1), and Sulfatase 1 (SULF1). β-actin (ACTB) was used as an endogenous control.
Total RNA was reverse transcribed into cDNA using the Ovation Pico WTA System (NuGen Technologies, San Carlos, CA). cDNA samples were cleaved and 3′-biotinylated using the Encore Biotin Module (NuGen Technologies, San Carlos, CA), yielding biotinylated single-stranded cDNA probes of 50–100 nt in length. The probe mix was hybridized with the GeneChip Human Genome HT U133 Plus PM Array plate (Affymetrix, Santa Clara, CA), presenting >54,000 target probe sets representing 47,000 transcripts encoding >33,000 well characterized human genes. Microarray analysis was conducted by the Beth Israel Deaconess Medical Center Genomics and Proteomics Center according to standard Affymetrix protocol, using a high throughput hybridization and scanning system. The quality of hybridized chips was assessed using Affymetrix guidelines based on PM mean, 3′ to 5′ ratios for beta-actin and GAPDH, and values for spike-in control transcripts. We also monitored sample reproducibility by chip- to-chip correlation and signal-to-noise ratio (SNR) methods for replicate arrays using the bioconductor package, arrayQualityMetrics
To obtain signal values, chips were further analyzed using the Robust Multichip Average (RMA) method in R using Bioconductor and associated packages. RMA performed background adjustment, quantile normalization and final summarization of 11 oligonucleotides per transcript using the median polish algorithm
Identification of functionally related genes differentially expressed in the profiles of EPS, or in the profiles of both EPS and PD as compared to Uremic, was by self-organizing map (SOM) analysis of the differentially expressed genes. We performed SOM clustering on transcript expression values using Pearson correlation coefficient-based distance metrics and a target of 40 groups. SOM allowed grouping of gene expression patterns into an imposed structure in which adjacent clusters are related, thereby identifying sets of genes that follow common expression patterns (transcription signatures) across different conditions
Ingenuity Pathway Analysis (IPA 7.0, Ingenuity Systems, Inc., Redwood City, CA) was used to identify pathways and interaction networks significantly affected by genes with expression changes specifically associated with EPS or with both EPS and PD. The knowledge base of this software consists of functions, pathways and network models derived by systematic exploration of the peer-reviewed scientific literature. IPA analysis calculates a P-value for each pathway according to the fit of user data to the IPA database, using a one-tailed Fisher exact test (
An IPA score [(−log(P value)]>2 indicates a probability >99% that the affected network was not generated at random, but reflects statistically (and biologically) meaningful relationships among a set of genes with specifically altered expression. The ability to rank networks based on relevance to the queried data sets allows network prioritization according to highest predicted impact on the disease process.
Deparaffinized, rehydrated tissue sections were incubated in Peroxidase Blocking Solution (S 2023, DAKO, Hamburg, Germany). Immunostaining was performed with a TechMate 500 Plus (DAKO), using a dextran-coated peroxidase-coupled polymer system (Dako REAL™ EnVision™ Detection Kit, Peroxidase/DAB+, Rabbit/Mouse, K 5007, DAKO). The primary antibody was rabbit monoclonal anti-human collagen 1 α 1 (AB292, Abcam, Cambridge, MA) diluted 1∶100 in DAKO S 2022 diluent. Skin tissue from a human non-diabetic leg served as a positive tissue control. Omission of primary antibody served as negative control. Adjacent tissue sections were stained with Hematoxylin and Eosin (H&E).
Semiquantitative analysis of Collagen 1 α 1 immunostaining was performed within defined areas from 0–100 µm and from 100–200 µm inward from the mesothelial tissue edges towards the adventitia, using ImageJ 1.46 (NIH). Color images recorded by an Olympus BX41 microscope equipped with an Olympus DP71 camera were converted to 32-bit black and white. Immunostaining detection threshold was zeroed at background staining intensity, and ranged from 0–255 arbitrary units. Measurements were recorded at three or more discreet locations within each specimen. Scaled intensity values in the region 0–100 µm from the mesothelial edge were divided by the intensity values in the adjacent region 100–200 µm from the edge. The mean intensity ratios for each group were normalized to that of the Uremic group, to which was assigned a relative intensity ratio value of 1.0.
Clinical data and laboratory values were compared by ANOVA (Sigmaplot 11.0, Systat Software Inc., San Jose, CA). Differences in results were considered statistically significant for P<0.05.
Tissue samples were available from 8 patients: 4 with encapsulating peritoneal sclerosis (EPS, 2 males and 2 females, of mean age 60.8 yrs and mean PD duration 74±34 months), 2 chronic peritoneal dialysis patients (PD, 1 male and 1 female, of mean age 68.5 yrs and mean PD duration 23 months), and 2 male Uremic patients of mean age 56.5 yrs, as yet undialyzed (
Clinical data | EPS (4) | PD (2) | Uremic (2) |
Age [a] | 60.75±12.42 | 68.50 | 56.50 |
Sex | 2 m, 2 f | 1 m, 1 f | 2 m |
Time on PD [months] | 74.00±33.52 | 23 | - |
Kt/V | 1.96±0.32 | 2.41 | - |
Bacterial peritonitis [# of episodes] | 2.25±2.63 | 0.50 | - |
24 h urine output [mL] | 275.00±246.64 | 440.00 | 2000.00 |
Nicotine abuse | 2/4 | 0/2 | 2/2 |
Arterial hypertension | 4/4 | 1/2 | 2/2 |
Diabetes of any kind | 1/4 | 2/2 | 2/2 |
PD fluids | 2 neutral/2 acidic | both neutral | - |
Icodextrin use | 4/4 | 2/2 | - |
|
|||
Hemoglobin [g/L] | 111.75±7.63 | 100.00 | 131.50 |
Leucocytes [×109/L] | 8.79±3.41 | 4.70 | 6.40 |
CRP [mg/dL] | 2.97±3.35 | 1.35 | 0.30 |
Creatinine [mg/dL] | 3.55±1.84 | 4.00 | 5.80 |
BUN [mg/dL] | 86.58±36.40 | 61.00 | 173.00 |
Calcium [mM/L] | 2.35±0.10 | 2.09 | 2.51 |
Phosphate [mM/L] | 1.45±0.34 | 1.51 | 1.45 |
Clinical data and laboratory values of (n) patients, collected before the surgeries that allowed collection of peritoneal biopsy samples. Values are shown as means ± s.e.m. for the EPS group and as means for the smaller PD and Uremic groups. Smoking history was not stratified by duration. Uremic group 24 h urine output was higher than that of the combined PD and EPS groups (p<0.001). Arterial hypertension was defined as resting arterial blood pressure ≥140/90 mmHg. Acidic PD solutions were lactate-buffered with pH 5.0–5.5. Neutral (multicomponent) solutions were of pH 6.5. Icodextrin status was listed as positive if used at any time during the course of PD.
EPS, Encapsulating peritoneal sclerosis; PD, Peritoneal dialysis; CrP, C reactive protein; BUN, blood urea nitrogen.
Unsupervised analysis was performed on ∼11,500 transcripts after preprocessing and normalization of high quality DNA expression data. By principal component analysis, we demonstrated that samples separated according to EPS status along primary component (PC) 1, which accounted for 42.2% of the variation between samples. The analysis revealed three distinct clusters of gene expression by principal component analysis (PCA) (
The first component with highest variance (42.2%) is on the X-axis separating Uremic and PD from EPS samples. The second highest (26.4%) is on the Y-axis depicting maximum variation between PD and Uremic samples. The Uremic, PD and EPS samples formed three separate clusters on the PCA plot.
Supervised analysis applying a 90% lower confidence bound of the fold change (LCB of FC), and expressed as corrected fold-change >2.0, revealed 531 genes differentially expressed between EPS and PD groups, 557 genes differentially expressed between EPS and Uremic groups, and 816 genes differentially expressed between PD and Uremic groups (
The Venn diagram of
When using gene clustering and self-organizing maps (SOM) to detect groups of differentially expressed genes with similar expression patterns, we arbitrarily drew 40 separate maps according to Pearson correlation coefficient-based distance metrics (
We performed canonical pathways enrichment analysis (IPA 7.0) to gain further insight into the functional pathways associated with genes dysregulated only in EPS (EPS signature), and with those dysregulated in both EPS and PD (Disease signature). EPS-associated genes were significantly over-represented (P value<0.05) in pathways related to immune response, inflammation and cytoskeleton signaling, including
Pathway analysis of the genes differentially expressed in both the EPS vs. Uremic and the PD vs. Uremic comparisons showed significant association with pathways involved in cell signaling, immune response and metabolism, including
To integrate a functional view of critical regulatory molecules associated with EPS, we performed interactive network analysis on the 228 genes specifically dysregulated in EPS tissue specimens (
To gain further insight into functional consequences of genes that are altered both in both EPS and PD as compared to the Uremic group, we again performed IPA interactive network analysis, identifying 8 different networks significantly affected in both EPS and PD compared to the Uremic control group with a significance score >20. Three networks related to inflammatory and immunological diseases were merged to identify critical regulatory genes (
Selected cases of differential gene expression were validated by qRT-PCR assays. β-actin was used as an endogenous “control” transcript. The data were analyzed using the 2−Δ
Collagen 1 α 1 (Col1a1), Fibronectin 1 (FN1), and thrombospondin 1 (THBS1) were highly upregulated in EPS compared to PD and Uremic groups, while retinol-binding protein 4 (RBP4) was highly downregulated in EPS and PD groups compared to the Uremic group.
To determine protein expression of Col1a1, one of the more highly upregulated genes in EPS, we performed anti-Col1a1 immunohistochemical staining of formalin-fixed, paraffin-embedded tissue sections corresponding to the fresh frozen specimens from which RNA was isolated. EPS and PD tissues showed marked immunostaining for Col1a1, indicating fibrosis. In most of the EPS and PD samples the mesothelium was not detectable. PD sections were strongly and homogeneously stained with Col1a1 throughout the entire specimen. In contrast, EPS samples exhibited pronounced Col1a1 immunostaining predominantly in the area of the (most superficial) submesothelial zone. We quantified the extent of the latter staining pattern by making staining intensity measurements in the submesothelial zone (0–100 µm) and the adjacent, deeper zone (100–200 µm from the surface of the tissue block zone moving towards the adventitial layer). The normalized submesothelial staining intensity of the 0–100 µm zone relative to the deeper 100–200 µm zone was 7-fold higher in the EPS group than in the PD or Uremic groups (
Peritoneal biopsy sections from (
EPS is among the most serious complications of chronic PD. Although PD duration seems to be a major risk factor, spontaneous EPS has been reported, and the causes of EPS remain obscure. The factors that govern progression from the common PD-associated condition of simple peritoneal fibrosis to the rare but aggressive condition of EPS are poorly understood. The pilot study presented here is the first comparison of transcriptomes of human peritoneal tissues taken from patients with EPS, PD patients without EPS, and predialytic uremic patients. The three groups show distinct gene expression patterns that allow their separation by principal component and heat map analyses. The results constitute proof-of-principle for a larger-scale study that could generate hypotheses for future EPS research, leading to more rigorous biochemical or cell biologic diagnostic criteria for EPS, and to definition of prognostic markers. Correlation of these and future transcriptome data with proteomics studies from peritoneal tissue and fluid would add further insight into EPS disease pathways and could improve early diagnosis and prevention.
The samples examined in this study were the first fresh-frozen tissue samples collected by the Peritoneal Biobank at the Robert-Bosch Hospital in Stuttgart, Germany, as part of a comprehensive database of tissues, serum, whole-blood and peritoneal effluent samples from PD patients, PD patients with EPS, and control patients without and with uremia. Collection of formalin-fixed, paraffin-embedded peritoneal tissue blocks had been initiated previously. The early stage of fresh tissue collection is reflected in the small sample sizes of the current study and the imperfect clinical matching of specimens in terms of PD duration, number of peritonitis episodes, Kt/V, systemic baseline inflammatory state, and use of acidic dialysate likely containing glucose oxidation products
The limited number and mass of patient samples available for this pilot study also prevented uniformity of dominant tissue histological features of the RNA source tissues within and among clinical groups. Later studies might employ laser capture microdissection to facilitate comparison of histologically similar regions within and among groups.
Transcriptome results exhibited good inter-individual agreement. Unsupervised analysis identified three distinct gene expression patterns consistent with the samples' clinical groupings (
The differentially expressed genes could be organized into canonical pathways that distinguished EPS as a separate group from both the PD and the Uremic groups (
The pathways shared by EPS and PD groups that differentiate them both from the Uremic group (
The myofibroblast marker α-smooth muscle actin (ACTA2) mRNA was also upregulated (
Several transcriptome studies of rodent peritoneal fibrosis models have recently appeared. Yokoi et al
Histochemical expression of collagen I was compared to the increased Col1a1 mRNA levels in EPS tissue. The increased Col1a1 mRNA level in EPS tissue was not paralleled by higher total Col1a1 polypeptide levels, as judged by immunostaining intensity across entire fields of view. Although EPS tissue showed greater Col1a1 expression than did Uremic group tissue, the highest levels of Col1a1 immunostaining were observed in PD tissues. In contrast to the homogeneous distribution of Col1a1 immunostaining across entire PD tissue sections, EPS samples exhibited pronounced intensification of Col1a1 immunostaining in the submesothelial compact zone (
Although EPS remains a poorly understood disease, we have demonstrated a distinct transcriptional pattern of peritoneal tissue from EPS patients that differed from those of PD patients and predialytic uremic patients. The distinct transcriptional patterns of EPS and PD tissues presented here are consistent with the hypothesis that EPS constitutes a separate disease entity, possibly requiring a “second hit”, and not merely an exacerbation of the simple peritoneal sclerosis believed present in all PD patients. However, the data also are consistent with EPS representing a malignant acceleration of PD-associated peritoneal fibrosis influenced by patient-specific risk modifier genes and, possibly, by transplant and/or immunosuppression therapy.
Although our small study does not point to a principal pathogenic mechanism in EPS distinct from that of the simple peritoneal fibrosis of chronic PD, the study does serve as proof-of-principle for future array studies with peritoneal tissue from larger groups of patients. Preliminary data (not shown) indicates that the spectrum and abundance of mRNA isolated from formalin-fixed paraffin-embedded tissues from the same patients studied in this paper were very similar to those of mRNA isolated from fresh-frozen tissue (correlation coefficient >0.91). This suggests the reliability of transcriptome analysis applied to archived paraffin blocks of tissues from EPS and PD patients, especially when combined with laser capture microdissection for comparison of histologically similar regions of tissue. Similar transcriptome studies should be conducted with peritoneal dialysate fluid and cell specimens, and with blood specimens, since accrual of samples will be easier than for peritoneal tissue samples. Moreover, definition of transcriptional signatures may prove to have value for diagnostic monitoring and prognosis. Comprehensive analysis of miRNAs will be important to include in these future studies, since they (or their antagomirs) can be easily envisioned as intraperitoneal therapeutic agents.
(DOC)
(DOC)
(DOC)
(DOC)
(TIF)
(TIF)
(TIF)
(TIF)
We thank Boris E. Shmukler, Augustine Rajakumar, Dongsheng Zhang, and Jack Lawler for helpful discussion and sharing reagents.