Introduction
Twenty years ago, the chronic bladder disease “interstitial cystitis” (IC) was widely regarded as an inflammatory condition. Published IC research in the early 1990s was based primarily on biopsy and histological investigations [
1,
2]. However, a debate began during this same time about the need for cystoscopy and biopsy in order to accurately diagnose IC [
3]. Patients presenting with clinical symptoms such as pain and urinary frequency, in the absence of any known cause, were commonly diagnosed with “IC syndrome” [
3]. Due to the subjectivity of this definition, a diverse group of patients were identified with IC ranging from those with neuropathic changes to those having a severe inflammation with macroscopically obvious degeneration of the urothelium. These broad inclusion criteria have made it difficult to establish diagnostic markers for the disease.
A consensus definition of “bladder pain syndrome/interstitial cystitis (BPS/IC)” was proposed by the European Society for the Study of Interstitial Cystitis (ESSIC). The diagnosis is based on chronic pelvic pain, pressure or discomfort perceived to be related to the urinary symptom accompanied by at least one other urinary symptom such as persistent urge to void or frequency [
4]. Confusable diseases [i.e. carcinoma, infection, overactive bladder (OAB)] need to be excluded. The severity of BPS/IC is classified by the rating of cystoscopic and histopathological evidence [
4]. The maximum cystoscopic severity is the finding of Hunner’s lesions, and the maximum histopathological severity is the finding of inflammatory infiltrates and/or detrusor mastocytosis and/or granulation tissue and/or intrafascicular fibrosis. These maximum findings are referred to as “ESSIC type 3C” [
4].
In an effort to more precisely define suitable BPS/IC diagnostic markers, our research team conducted a pilot study in 2009 comparing gene expression analysis (Affymetrix GeneChip® Human Genome U133 Plus 2.0) of bladder biopsies from five patients with BPS/IC ESSIC type 3C and six healthy controls [
5]. Approximately 1,000 genes from more than 50,000 probe sets were found to have a BPS/IC to healthy expression ratio greater than two. The corresponding gene expression patterns were characteristic for immune system diseases, and they predicted chronic inflammation with B- and T-lymphocyte infiltration. In addition, the array data suggest an abnormal differentiation of the urothelium which is in agreement with earlier results [
6‐
8]. Two urothelium-specific genes, cytokeratin 20 (
KRT20) and uroplakin 1B (
UPK1B), were found to be expressed at very low levels [
5].
While gene expression arrays can provide a large amount of information, the process is expensive, time-consuming and requires specialized equipment. We investigated if it was possible to characterize Hunner’s positive BPS/IC by the quantification of seven representative urothelial, B- and T-cell markers. We hypothesized that a local B-cell activation would lead to high levels of antibody concentration in urine.
Materials and methods
Description of patients
This prospective cohort study assigned patients recruited at one urogynaecological centre in Switzerland from 2006 to 2011 to three comparison groups: BPS/IC ESSIC type 3C (group 1, n = 15), OAB (group 2, n = 11) and healthy controls (group 3, n = 8). Each woman reviewed and completed relevant study documentation approved by the local ethics commission including an informed consent form (reference 9900.013).
In order to detect markers, homogeneous groups of patients were required. Consequently, patients with predominant pain symptoms were selected based on their diagnosis with classic ulcerative IC or IC with Hunner’s lesions that conformed to the BPS/IC ESSIC type 3C criteria [
4]. “Hunner’s lesions” were identified according to ESSIC [
4] as distinctive, circumscript reddened mucosal areas with small vessels radiating towards a central scar that, upon bladder distension, will rupture and bleed. As a control group, we chose patients who had primary urgency symptoms and a cystoscopically non-inflamed bladder wall (“non-inflammatory OAB”). These patients represent a subgroup of the OAB condition since there are also reports of an association between OAB and inflammation [
9‐
11]. Healthy control patients were identified as women with no bladder symptoms and who were hospitalized for a hysterectomy shortly before enrolment in the study. Patients with urinary tract infections were excluded.
Biopsy removal and real-time quantitative polymerase chain reaction (RT-qPCR)
Two cold-cup biopsies were taken from each patient and, when present, from the transition zone lesion/non-lesion. Thus, bladder biopsies included various amounts of urothelium, submucosa and muscularis propria (detrusor). Histopathological testing of one of these tissue samples included mast cell quantification. Detrusor mastocytosis was defined according to ESSIC [
4] as mast cell counts greater than 28 mast cells/mm
2. The other biopsied tissue was preserved in RNAlater (76104, QIAGEN, Valencia, CA, USA) and used for RNA isolation and quantification as described in the report of our pilot study [
5]. Reverse transcription and RT-qPCR were done with materials and equipment from Applied Biosystems (Life Technologies, Zug, Switzerland) as follows: TaqMan® Reverse Transcription Reagents (N8080234) for reverse transcription, inventoried Human TaqMan® Gene Expression Assays (4331182) and TaqMan® Universal PCR Master Mix (4364340) for RT-qPCR on an Applied Biosystems 7500 Real-Time PCR System. The following TaqMan® Gene Expression Assays (4331182) were analysed: cytotoxic T-lymphocyte-associated protein 4 (
CTLA4, Hs03044418_m1) as a T-cell marker;
CD20 (membrane-spanning 4-domains, subfamily A, member 1
MS4A1, Hs00544819_m1) and
CD79A molecule, immunoglobulin-associated alpha (
CD79A, Hs00233566_m1) as B-cell markers; immunoglobulin heavy locus (
IGH@, Hs00378230_g1) as a marker for antibody heavy chain gene expression; and uroplakin 1B (
UPK1B, Hs00199583_m1), uroplakin 3A (
UPK3A, Hs00199590_m1) and cytokeratin 20 (
KRT20, Hs00300643_m1) as urothelial markers. RT-qPCR reactions were done in duplicates in a 25 μl volume. Standard cycling conditions were used (2 min 50 °C, 10 min 95 °C, 40 cycles 15 s 95 °C/1 min 60 °C). Relative gene expression (comparative Ct method quantitation) was calculated with the 7500 System SDS software version 1.3. Results were given as “relative expression” using glyceraldehyde-3-phosphate dehydrogenase (
GAPDH, Hs99999905_m1) as the endogenous control and healthy control patient 005 [
5] as the calibrator.
Immunohistochemistry
As described in our previous study, biopsied tissue was embedded in paraffin and sections were stained with either haematoxylin and eosin or Giemsa and van Gieson elastin [
5]. For immunohistochemistry testing, biopsy sections were treated according to the standard protocol with the Leica BOND-MAX automated system using Leica Novocastra reagents (Biosystems, Nunningen, Switzerland). The standard protocol for immunohistochemistry was as follows: Dewaxing (AR9222), Epitope Retrieval Solution 2 (AR9640) and Bond Polymer Refine Detection Kit (DS9800) or Bond Polymer Refine Red Detection Kit (DS9390). The primary antibodies were 1F6 mouse monoclonal antibody to cluster of differentiation 4 (CD4, NCL-CD4-1F6, Novocastra) diluted 1:50; MJ1 mouse monoclonal antibody to CD20 (NCL-CD20-MJ1, Novocastra) diluted 1:100; 11E3 mouse monoclonal antibody to CD79A (NCL-CD79a-225, Novocastra) diluted 1:125; AU-1 mouse monoclonal antibody to uroplakin 3 (UPK3, 345 M-14, Cell Marque, Rocklin, CA, USA) diluted 1:25; and PW31 mouse monoclonal antibody to KRT20 (NCL-L-CK20-561, Novocastra) diluted 1:200. Microphotographs were taken at ×20 magnification with a UC30 colour camera mounted on a BX43 Olympus System Microscope. Images were processed by the CellF image analysis software (Olympus, Volketswil, Switzerland).
Blood handling and quantification of plasma IgG and IgA
Blood was collected in VACUETTE® Heparin tubes (Greiner Bio-One, St. Gallen, Switzerland) one day before biopsy. IgG and IgA concentrations were determined on a cobas® 6000 analyzer (Roche Diagnostics AG, Rotkreuz, Switzerland).
Urine handling and quantification of urinary immunoglobulins
Urine was obtained through urinary catheterization one day before the biopsy. Standard laboratory techniques included a routine urine status and a urine culture. Sterile urine was immediately centrifuged (5 min, 900 g), and the supernatant was frozen at −20 °C and subsequently stored at −80 °C. After thawing, protease inhibitor (Complete 04693116001, Roche Diagnostics AG, Rotkreuz, Switzerland) was added and 300 μl-aliquots were stored at −80 °C. Concentration of urine creatinine was determined with the CREA plus kit from Roche/Hitachi on a Roche/Hitachi Modular System P Chemistry Analyzer (Roche Diagnostics AG, Rotkreuz, Switzerland) at Ilamed AG, Institute for Laboratory Medicine (Frauenfeld, Switzerland). For immunoglobulin quantification, urine aliquots were thawed and centrifuged for 15 min at 2,000 g. Supernatants were adjusted to pH 7.4 by adding a modified phosphate-buffered saline (PBS) buffer (final 1× concentration: 137 mM NaCl, 2.7 mM KCl, 40 mM phosphate buffer, pH 7.4). To quantify the protein concentration, three different dilutions were measured per urine sample. IgG and IgA concentrations were determined on a BD FACSCalibur flow cytometry instrument using Human Total IgG CBA Flex Set (No. 558679) and Human Total IgA CBA Flex Set (No. 558681) kits in accordance with the manufacturer’s protocol (BD Biosciences, Allschwil, Switzerland).
Statistics
Statistical analyses were done in IBM® SPSS® Statistics (version 19). A
p value of less than 0.05 was considered to indicate statistical significance. Tests of normality (Shapiro-Wilk) and homogeneity of variance (Levene statistic) were done with all continuous variables. The Kruskal-Wallis test was used for non-parametric data. Post hoc tests (Mann-Whitney U) identified differences between two groups, and the Bonferroni correction was applied. As such, a
p value of 0.016 (0.05/3) indicated significance. For normally distributed data, analysis of variance (ANOVA) followed by Tukey’s HSD post hoc tests were used to detect differences between the groups. Non-parametric receiver-operating characteristic (ROC) curves were generated for all biomarkers to plot the sensitivity against the false-positive rate (1-specificity). To accomplish this, biomarker values for BPS/IC ESSIC type 3C were compared to values of OAB combined with the healthy control group. The optimal cut-off for each biomarker was selected to maximize the sum of sensitivity and specificity [Youden index = max. (selectivity + specificity-1] [
12]. The accuracy of a biomarker to predict BPS/IC, defined as the average of sensitivity and specificity, was also calculated.
Discussion
Gene expression analysis of four T- and B-cell markers (CTLA4, CD20, CD79A and IGH@) and three urothelial markers (UPK1B, UPK3A and KRT20) revealed significant differences between the BPS/IC ESSIC type 3C group and the two control groups. High concentrations of IgG and IgA in urine, but not in blood, were associated with high expression of lymphocyte-specific genes and follicular B- and T-cell aggregates in the submucosa and were consistent with our hypothesis that there is a local immune response with B-cell activation in cases of BPS/IC ESSIC type 3C. In addition, urothelium-specific gene expressions were low in BPS/IC ESSIC type 3C. One urothelial marker (UPK1B) even reached significantly lower expression for non-inflammatory OAB when compared to the healthy control group. For most other gene expression markers and the urine immunoglobulin concentrations, the OAB values lie between those of the healthy control group and the severe BPS/IC form.
The results for BPS/IC ESSIC type 3C are consistent with earlier findings that describe severe inflammatory infiltrates and an abnormal urothelium for this disease subtype [
1,
2,
6‐
8,
13‐
16]. In contrast to previous studies, though, we provided a subset of quantifiable in vivo gene expression markers for both the inflammatory and the urothelial aspects of the disease. Measurement of these markers may aid in the detection of BPS/IC ESSIC type 3C.
Although urinary immunoglobulins have been detected in earlier studies [
17], this is the first report on BPS/IC ESSIC type 3C to show significantly high levels of antibodies in urine.
OAB and BPS/IC are regarded as different syndromes—
urgency is considered the hallmark symptom of OAB and
pain is the hallmark symptom of BPS/IC [
18]. In our study, OAB patients with non-inflammatory cystoscopic findings served as a control group. Our intention was to determine if the marker values for the OAB group were more similar to those of the BPS/IC group, the healthy group, or if the values were at intermediate ranges between these two groups since other recently published studies found an overlap between OAB and BPS/IC on both a clinical and diagnostic level [
9‐
11,
18]. Under this assumption, the results we obtained for the OAB group can be explained.
The low expression of the urothelial genes in BPS/IC type 3C might be due to a loss of urothelium within the Hunner’s lesion, which has also been described by other researchers [
19,
20]. However, in our previous study [
5], two biopsies of a Hunner’s positive BPS/IC bladder were analysed—one from the transition zone Hunner’s lesion/non-lesion and the second from a normal-looking area of the bladder. The urothelial genes
UPK1B (210064_s_at) and
KRT20 (213953_at) were shown to be expressed at significantly lower levels, even in the non-affected, normal-looking bladder tissue [
5].
Only a few studies addressed the impact of urothelium integrity on OAB; however, Liu et al. [
10] recently found that the urothelial barrier function was normal in OAB patients. Conversely, findings from our study showed significantly low expression of one urothelial marker (
UPK1B) in non-inflammatory OAB patients compared to the healthy control group. Limited comparisons can be made between these two studies since Liu et al. used patients with stress urinary incontinence as controls rather than patients free of bladder symptoms. Moreover, they performed integrated image sections measuring immunofluorescence rather than comprehensive quantitative gene expressions, and they looked at different urothelial markers (zonula occludens-1 and E-cadherin rather than
UPK1B,
UPK3A and
KRT20).
Even though efforts have been made [
13,
21], there are currently no commonly accepted markers to diagnose BPS/IC. Furthermore, there are only invasive techniques to differentiate Hunner’s positive from Hunner’s negative BPS/IC. It is likely that the IgA and IgG urine findings are due to the presence of lymphocyte infiltrates in the submucosa and would not be elevated if the histology were normal. As part of our future research, we plan to investigate whether elevated urinary IgA and IgG levels are indicative of Hunner’s lesion, which would mean a non-invasive technique to identify this disease subtype would be available. Since an association between BPS/IC and autoimmune diseases has been suggested [
22], determining the specificity of the urinary immunoglobulins would likely give new information on the catalyst and the pathophysiology of this local immune reaction.
The B-cell markers
CD79A and
CD20 are similarly expressed (Table
2); therefore, quantification of one of these representative markers should be sufficient. The elevated number of B-cells in the bladder wall is an important consideration for prospective therapeutic approaches to BPS/IC ESSIC type 3C. Rituximab, a monoclonal antibody specific for CD20, selectively destroys B-cells [
23]. Given our findings of high
CD20 expression in Hunner’s positive BPS/IC, we would propose treatment with rituximab, a treatment which currently is approved for use with non-Hodgkin’s lymphoma, rheumatoid arthritis and ANCA (anti-neutrophil cytoplasmic antibodies)-associated vasculitis [
23,
24]. By tracking changes in gene expression, we may be able to quantify and monitor therapy progress and outcomes.
With regard to study limitations, the low number of study participants was largely a consequence of a small pool of eligible patients at one recruitment centre and a relatively rare condition. Furthermore, the healthy controls were patients hospitalized for gynaecological surgery and underwent cystoscopy and bladder biopsy solely for study purposes. For this reason, patient groups could not be matched by age, which could signify a potential source of bias. Similar to other studies [
25,
26], patients with Hunner’s lesions were older than the comparison groups.
We chose to exclude patients with Hunner’s negative BPS/IC since they make up a large and heterogeneous group. This study was intended as an initial investigative step to find biomarkers; therefore, we focused our research efforts on the small homogeneous group of patients with Hunner’s positive BPS/IC who could easily be identified with cystoscopy. For future studies, a careful subdivision of the heterogeneous Hunner’s negative BPS/IC group based on the ESSIC grading system (i.e. cystoscopy and biopsy data) would be required.
Tests of sensitivity and specificity showed that all markers were accurate biomarkers for BPS/IC ESSIC type 3C (Table
3). However, these results were obtained from small, homogeneous subgroups of patients. Further analyses are planned with a larger cohort of patients, including those with Hunner’s negative BPS/IC.
In summary, this study identified seven quantifiable gene expression markers which can describe the current stage of the clinically defined subtype of BPS/IC ESSIC type 3C. Objective, measurable results can be generated within 2 days and more focused monitoring of therapy outcomes could be carried out. Further research is needed to develop more effective strategies for treating Hunner’s positive BPS/IC, such as targeted therapy with monoclonal antibodies against B-cells. Detection of elevated urinary IgA and IgG levels could be a non-invasive technique to diagnose the subtype BPS/IC ESSIC type 3C representing ∼10 % of all BPS/IC patients [
20]. Lastly, since the results for the non-inflammatory OAB group were in intermediate ranges between BPS/IC ESSIC type 3C and the healthy control groups, future research efforts ought to assess if BPS/IC and OAB are related.