Parallel changes in bladder suburothelial vanilloid receptor TRPV1 and pan-neuronal marker PGP9.5 immunoreactivity in patients with neurogenic detrusor overactivity after intravesical resiniferatoxin treatment

Authors


Prof. P. Anand, Imperial College London, Peripheral Neuropathy Unit, Hammersmith Hospital, Du Cane Road, London W12 0NN, UK.
e-mail: p.anand@imperial.ac.uk

Abstract

OBJECTIVE

To compare PGP9.5 and transient receptor potential vanilloid receptor (TRPV1) suburothelial immunoreactivity between controls and patients with spinal neurogenic detrusor overactivity (NDO) before and after treatment with intravesical resiniferatoxin, as suburothelial PGP9.5-staining nerve fibres decrease in patients with spinal NDO who respond to intravesical capsaicin, and TRPV1 is present on these suburothelial nerve fibres in normal and overactive human urinary bladder.

PATIENTS AND METHODS

Patients with refractory NDO were enrolled in a prospective, randomized, parallel-group, double-blind, placebo-controlled trial using escalating doses of resiniferatoxin to a maximum of 1 µmol/L. Flexible cystoscopic bladder biopsies obtained at baseline, 4 weeks after each instillation and at the time of maximum clinical response were compared with biopsies taken from control subjects. Frozen sections were incubated with rabbit antibodies to TRPV1 and PGP9.5, and assessed using standard immunohistochemical methods. PGP9.5 nerve density was analysed using a nerve-counting graticule by an observer unaware of sample origin. Another two independent observers unaware of each other's results used a random grading scale to evaluate TRPV1 nerve fibre density and intensity. The immunohistochemistry results were compared with histology findings (haematoxylin-eosin), and the Mann–Whitney test used to assess any differences (P < 0.05 significant) and the Pearson test for correlation.

RESULTS

There were eight controls and 20 patients with spinal NDO, 14 (five clinical responders and nine not) who received the maximum dose of resiniferatoxin. There were more PGP9.5 and TRPV1 nerve fibres in patients with NDO than in controls (P = 0.007 and 0.002, respectively). Immunoreactivity before resiniferatoxin was similar in both groups for both PGP9.5 and TRPV1. In responders there were fewer PGP9.5 and TRPV1-positive fibres after treatment (P = 0.008 for each) but no change in those not responding. Changes after treatment for TRPV1 correlated well with those for PGP9.5 (r = 0.88, < 0.001).

CONCLUSIONS

The decrease of PGP9.5 and TRPV1 immunoreactive nerve fibres in responders to resiniferatoxin (to levels in control tissues) suggests that the increased numbers of nerve fibres in patients with NDO are mainly of sensory origin and express TRPV1. As baseline nerve fibre values were similar in responders and nonresponders, an additional factor may account for the difference in treatment outcome.

Abbreviations
NDO

neurogenic detrusor overactivity

CGRP

calcitonin gene-related peptide

DSD

detrusor sphincter dyssynergia

PVC

polyvinyl chloride

MS

multiple sclerosis

NGF

nerve growth factor.

INTRODUCTION

Neurogenic detrusor overactivity (NDO) in animal models is associated with the emergence of an aberrant segmental sacral spinal reflex that is mediated by vanilloid-sensitive C-fibre afferents in the bladder suburothelium [1]. In patients with spinal NDO, treatment with intravesical vanilloids such as capsaicin or resiniferatoxin results in increased bladder capacity, and decreased urinary frequency and urge incontinence episodes [2–4]. Unfortunately the reversible suppression of sensory neurone activity (desensitization) that follows the instillation of 1–2 mmol/L capsaicin is preceded by several days of bladder pain and overactivity associated with the acute release of neuropeptides (sensitization). In contrast, resiniferatoxin induces desensitization at lower concentrations (10 nmol/L [5], 50 and 100 nmol/L [3,4]) than capsaicin with minimal noxious effects [3–5].

The action of vanilloid compounds is mediated via the vanilloid receptor ‘transient receptor potential V1’ (TRPV1; also known as VR1), a member of the TRP protein superfamily, consisting of a diverse group of Ca++ permeable, nonselective cation channels that mediate the influx of extracellular Ca++ in response to depletion of intracellular Ca++ stores [6,7]. TRPV1 is a 95-kDa capsaicin-gated cation channel [8,9] expressed by primary sensory C-fibres in animal and human suburothelium [10–12]. Mice lacking the TRPV1 receptor have impaired detection of painful heat and do not show vanilloid-evoked pain behaviour [13]. In addition, TRPV1 appears to be involved in regulating normal lower urinary tract function, as animals that lack the receptor have greater short-term voluntary urination and abnormal urodynamic responses, with an increase in the frequency of non-voiding contractions, an increased bladder capacity and inefficient voiding [14]. Bladder histology, urothelial cell ultrastructure and density of substance P- or calcitonin gene-related peptide (CGRP)-containing suburothelial nerve fibres were no different between control and TRPV1 ‘knockout’ mice. It was concluded that TRPV1 is involved in detecting bladder stretch in the normal animal [14].

PGP9.5 is a 212-amino acid protein present in tissues of human neuronal and neuroendocrine origin [15,16], and immunohistochemistry using anti-PGP9.5 is particularly useful for identifying small-diameter unmyelinated neurones [17]. We previously showed that the density of suburothelial nerves immunostained for the pan-neuronal marker PGP9.5 is greater in patients with spinal NDO than in controls, and that the mean suburothelial nerve density decreases in patients who respond to capsaicin [18]. We also showed that TRPV1-immunoreactive fibres were significantly less common than those that immunostained for PGP9.5 in cystoscopic biopsies from both normal postmortem and spinal neurogenic bladders, raising the possibility of differential expression of receptors in subsets of nerves in the suburothelium [10]. The aim of the current study was to examine suburothelial nerve fibre PGP9.5- and TRPV1-immunoreactive in flexible cystoscopic biopsies taken from patients with NDO before and after treatment with intravesical resiniferatoxin, and to compare these with biopsies from control subjects.

PATIENTS AND METHODS

Bladder biopsies were obtained from patients with refractory spinal NDO with or without detrusor sphincter dyssynergia (DSD) recruited into a prospective, randomized, parallel-group, double-blind, placebo-controlled trial to evaluate the safety, tolerance and efficacy of three concentrations of intravesical resiniferatoxin (RTX107; sponsor, Afferon Corp., Wayne, PA, USA) and placebo. Non-neurogenic control specimens were obtained from patients under investigation for asymptomatic microscopic haematuria. Anticholinergic medications were withdrawn for 10 days and a sterile MSU specimen confirmed before accrual of baseline urinary assessments that included frequency-volume charts and cystometry. Trial patients were randomized to receive an intravesical instillation of placebo, 0.1, 0.5 or 1 µmol/L resiniferatoxin. Safety, tolerance and efficacy were assessed immediately after dosing (safety assessments only) and at 1, 4, 8 and 12 weeks and monthly thereafter, or until there was a return to bladder function before treatment. Patients who responded to resiniferatoxin were re-treated with the next highest dose when the symptoms returned to baseline. Similarly, those not responding were re-treated until a response was obtained or at most two treatments of 1 µmol/L had been given.

Preparation and administration of resiniferatoxin

Sterile stock solution (1, 5 or 10 mL; 10 µmol/L or 6.3 µg/mL resiniferatoxin in 100% dehydrated alcohol stored at −10 to − 20 °C) was diluted in 9, 5 or 0 mL of dehydrated alcohol and 90 mL sterile 0.9% saline to achieve dosing concentrations of 62.9 (0.1), 314.4 (0.5) or 628.7 ng/mL (1.0 µmol/L) resiniferatoxin. The solutions were prepared in polyvinyl chloride (PVC)-containing sterile containers which were then refrigerated and stored for up to 24 h before administration. The time of preparation and instillation of resiniferatoxin were recorded.

Up to 100 mL resiniferatoxin or placebo was instilled into an emptied bladder at ≈50 mL/min and retained in the bladder for up to 30 min. It was not known at the time of the study that resiniferatoxin adsorbs to polypropylene and PVC present in the fluid bags, giving sets, syringes and catheters used in the study, with 20% of the resiniferatoxin bound by 30 min, 50% by 2 h and 80% by 4 h and up to 24 h at room temperature or 4 °C across a range of doses. Therefore the dose of resiniferatoxin delivered was retrospectively estimated according to resiniferatoxin degradation curves (Table 1; personal communication, ICOS Corporation, Bothell, Washington, USA).

Table 1.  Incubation time and corresponding doses (µmol/L) of resiniferatoxin received per patient for groups responding or not
NResponding (five)Not responding (nine)
Incubation, hdoseIncubation, hdose
1  0.750.72.00.5
2  2.50.43.50.3
322.50.22.00.5
4  2.50.41.50.55
5  1.50.554.750.2
6  1.750.55
7  2.00.5
8  3.00.35
9  1.50.55
Mean  5.950.452.440.44
Median  2.50.442.00.50

Bladder biopsies

Flexible cystoscopic bladder biopsies were taken as previously described from a consistent site, just above and lateral to the ureteric orifices [19], at baseline and 4 weeks after the first and each subsequent instillation of resiniferatoxin. In responders biopsies were also taken at the time of continuing clinical response. A urine specimen was sent for culture before each cystoscopy. Approval by the local ethics committee and informed consent were obtained from patients and controls.

Preparation and staining of specimens

For haematoxylin and eosin staining, each specimen was immediately fixed in freshly prepared 4% paraformaldehyde (w/v) in PBS (0.05 mol/L phosphate, 0.9% w/v saline, pH 7.2) for 30–60 min and then transferred to 0.45 mol/L sucrose in PBS and refrigerated overnight. The samples were embedded in OCT medium (Ames Corp., Iowa, USA) and 10 µm sections cut using a cryostat and stained with haematoxylin and eosin.

For PGP9.5 immunostaining [18], several 10 µm sections fixed in 4% paraformaldehyde as described above and mounted on slides were stored at − 60 °C before staining. After thawing the sections were soaked in 100 µL polyclonal antiserum to PGP9.5 per slide. The slides were then incubated with primary antibody (anti-PGP antibody at 1 : 1000 dilution; Ultraclone Ltd, UK) in a high-humidity chamber for 18 h. After three washes in PBS the specimens were further incubated with secondary antibody, anti-IgG biotinylated species-specific whole antibody at 1 : 1250 dilution (Amersham, UK) for 1 h. After three further washes with PBS they were incubated with streptavidin fluorescein at 1 : 100 dilution (Amersham) for 1 h. Three further washes were in PBS were followed by staining with pontamine sky blue (0.1% in 1% DMSO/PBS) for 2 min. Finally the slides were washed twice further with PBS and mounted using Cityfluor (Agar Scientific Ltd, UK).

For TRPV1 immunostaining, an affinity-purified rabbit antibody (♯C22, GlaxoSmithKline, UK) raised against a synthetic peptide sequence of human TRPV1 sequence was used. The specimens were immediately fixed as described above and stored at 4 °C; 4–5 µm frozen sections were collected onto clean glass slides. After washing in PBS, endogenous peroxidase was blocked by incubation with 0.3% w/v hydrogen peroxide in methanol. After a further wash in PBS, the tissue sections were incubated with rabbit antibody to human TRPV1 (1 : 5000) overnight. Sites of antibody attachment were revealed using a nickel-enhanced avidin-biotin (peroxidase) method [20]. Nuclei were counterstained with 0.1% w/v aqueous neutral red.

Analysis

Two pathologists (T.J. and A.F.) unaware of both the origin of the tissue (controls or NDO, before or after resiniferatoxin or placebo) and the clinical response to treatment assessed each section for inflammatory changes, site of inflammation, vascularity and urothelial changes (hyperplasia and dysplasia). Where inflammation (acute or chronic) was present, a subjective score was used that ranged from 1 (mild) to 3 (severe). The histological findings were correlated with the results of MSU and immunohistochemistry.

For PGP9.5 nerve density analysis a nerve-counting graticule was used superimposed over a × 200 field of vision. The 10 × 10 graticule grid covered an area of 0.495 mm2 of the specimen. Counts were made on three sections of tissue per specimen; the total number of nerve profiles in 300 graticule squares was recorded. The observer (M.H.) was unaware of both the origin of the tissue and the clinical response to treatment.

A qualitative method of analysis was used to evaluate TRPV1 staining, with a random grading scale of nerve-fibre density and intensity of 0–3, increasing by increments of 0.5, each increment being equivalent to up to five additional immunoreactive fibres per section. At least three and up to 12 sections were analysed for each evaluated biopsy and results expressed as mean of two independent observers (A.A. and Y.Y.) unaware of each other's results.

The Mann–Whitney test was used for the statistical analysis, with P < 0.05 considered to indicate statistical significance, and the Pearson test applied for correlation.

Twenty-two patients (13 women and nine men) with refractory spinal NDO were recruited into the study between October 1999 and June 2000. Thirteen patients had multiple sclerosis (MS), three had spinal cord vascular accidents, three had idiopathic inflammatory changes of the spinal cord (transverse myelitis), one had a history of cervical (C3/4) and L5/S1 intervertebral disc herniation, one a history of C5/6 intervertebral disc herniation with an ischaemic cervical cord (C2-6) and one had tropical spastic paraparesis.

RESULTS

Twenty patients had one or more treatments with varying doses of resiniferatoxin and 17 proceeded through the protocol to receive one or more instillations of the maximum permitted dose (label claim 1 mmol/L). Six of these 17 patients responded clinically to the treatment, showing marked improvements on frequency-volume charts and/or cystometry.

Flexible cystoscopic biopsies were available from eight controls (mean age 50.6 years, sd 21.5), from 20 patients (mean age 46.2 years, sd 10.6) with spinal NDO taken before resiniferatoxin treatment (‘baseline’) and from 14 taken after the maximum dose. Of these 14 patients five were responders who had sustained improvements on the frequency-volume chart and maximum cystometric capacity (three with MS, and one each with spinal cord vascular accident and idiopathic inflammatory change; mean age 45.6 years, sd 11.3; mean duration of disease 6.4 years, sd 3.4) and nine did not respond (seven with MS, one each with a spinal cord vascular accident and idiopathic inflammatory changes; mean age 47.9 years, sd 10.4; mean duration of disease 11 years, sd 10.9). There was a large but not statistically significant difference in the duration of NDO between responders and nonresponders. Biopsies from two of the five responders were also taken at the time of clinical relapse.

Four of five responders and six of nine nonresponders had evidence of DSD on baseline cystometry. There was no difference in the Kurtzke scores and duration of disease for the two response groups with MS. All patients had sterile urine cultures at the time of biopsy except one in each of the response groups and three treated with placebo who had UTIs at baseline biopsy.

The mean (median) number of treatments leading to the maximum dose was 2.6 (3) in the responders and 2.3 (3) in those not responding. Based on the incubation times before instillation the mean (median) estimated doses of resiniferatoxin and incubation times for both groups and the controls are shown in Table 1.

Histopathology

  • On haematoxylin-eosin staining, urothelium was present in all specimens and 54% contained muscularis propria. Mild urothelial hyperplasia was identified in two biopsies only (baseline, responders). Many cystic von Brunn nests were identified in the baseline and subsequent biopsies of one responder. There was increased vascularity in only one biopsy (baseline nonresponder) from a patient who had a UTI. The results of histopathological assessments relating to inflammatory changes are summarized in Table 2. The mean (sd) inflammation score was significantly higher in patients with NDO than in controls, at 1.45 (0.18) vs 0.62 (0.18) (P = 0.02). There was a marked but not significant difference (P = 0.06) between nonresponder baseline scores and control values. There was no evidence of dysplasia in any of the biopsies.

Table 2.  Incidence of inflammation in bladder biopsies from controls and patients with NDO before and after treatment with the maximum resiniferatoxin dose (R, response)
VariableControlsBeforeAfter
RNo RRNo R
Proportion showing inflammation5/84/58/93/54/9
Mean inflammatory score (0–3)0.61.41.41.00.8
Proportion showing acute inflammation1/81/51/90/51/9
Median (mean) time to biopsy after treatment, weeks6 (15)5 (4.6)

The mean density of PGP9.5-stained nerves in the suburothelium was significantly greater for the 20 baseline NDO biopsies than in the eight control biopsies (P = 0.007). The mean values before treatment were no different between the response groups. In the responders there was a significant decrease in PGP9.5-positive nerve density (P = 0.008; Table 3). In the two responders with available biopsies taken at the time of clinical relapse PGP9.5-immunoreactivity increased to pretreatment levels, being 148.3 and 213.7/mm2 before resiniferatoxin, 47.6 and 0.01/mm2 during the clinical response, and 276.8 and 250.2/mm2 at clinical relapse.

Table 3.  The mean (sem) nerve density of PGP9.5 and TRPV1 in the various groups before and after treatment with resiniferatoxin
GroupPGP9.5 positiveTRPV1+ve
BeforeAfterPBeforeAfterP
  • *

    comparison between NDO and controls.

NDO286.7 (45.5) 1.086 (0.17) 
Controls  74.2 (32.7)0.007*0.339 (0.17)0.002*
Responders229.1 (64.2)  23.0 (9.5)0.0081.594 (0.37)0.175 (0.07)0.008
Non-responders251.4 (77.1)284.0 (64.3) 1.030 (0.26)0.679 (0.16) 
Placebo-treated303.7 (71.5)348.5 (143.4) 0.881 (0.23)0.832 (0.26) 

Nerve densities did not change significantly in those not responding after treatment, with a mean density before and after treatment of 251.2 (231.4, 77.1) and 284 (193, 64.3), and in four patients after placebo, with respective values of 303.7 (143, 71.5) and 348.5 (286.7, 143.4).

For TRPV1, similar mean numbers of sections per biopsy were analysed for the NDO and control groups, at 6.40 (0.36, 4–10) and 6.87 (0.81, 5–12), respectively, and for responders and nonresponders, at 6.00 (0.55, 4–7) and 5.89 (0.42, 4–8), respectively. TRPV1-immunoreactive fibre-like structures were scattered throughout the suburothelium of both NDO and control samples, as previously described [10], but no interstitial cell staining for TRPV1 was identified, as reported elsewhere [12]. There were significantly more TRPV1-immunoreactive nerve fibres (Fig. 1) in patients with NDO than in controls (P = 0.002; Table 3). There was no difference in baseline values of TRPV1-immunoreactive fibres between the response groups, but in the five patients who responded, there were significantly fewer TRPV1-positive fibres (P = 0.008) after treatment than before, and similar to the controls (Table 3). In the two responders with available biopsies taken at the time of clinical relapse TRPV1-immunoreactive fibre density remained relatively lower than controls. There was no significant change in the density of TRPV1-immunoreactive nerve fibres in nonresponders after treatment and in the four controls after placebo therapy. Changes in TRPV1 after treatment correlated well with those in PGP9.5 in all patients (r = 0.88, P < 0.001).

Figure 1.

Fine-calibre, TRPV1-immunoreactive nerve fibres in the suburothelium of a patient with neurogenic detrusor overactivity (arrows indicate fibres). Uro, urothelium; Sub, suburothelium.

DISCUSSION

The therapeutic effect of resiniferatoxin in patients with refractory NDO caused by spinal cord disease has been reported in several studies, showing significant increases in bladder capacity, and decreases in urinary frequency and urge incontinence [3,4]. Patients treated with intravesical resiniferatoxin have minimal or no sensitization effects. However, as our clinical trial of intravesical resiniferatoxin progressed, several patients reported variable sensitization effects inconsistent with the ‘escalating dose’ design of the study. In addition, whilst the maximum dose used in this study was higher than in previously reported trials [3–5], surprisingly few patients had a clinical response (six of 17 who received a label claim of 1 µmol/L resiniferatoxin). These factors prompted an investigation into the stability of resiniferatoxin in the polypropylene and PVC-containing saline bags used during preparation of the drug. To ensure that the different clinical and immunohistochemical responses observed were not dose-related, we extrapolated the dose received by each patient based on the pre-instillation incubation time. In addition, repeated instillations with escalating doses might have had a cumulative effect; however, the total number of treatments received by each patient group up to and beyond the point of biopsy was also similar. Finally, patients with MS and significant lower limb dysfunction appear less likely to respond to intravesical vanilloid therapy [2]. However, in the present study the disease type did not correlate with response to treatment and, in patients with MS, the degree of lower limb disability as measured by the Kurtzke score was similar in responders and nonresponders, although there were few of either.

There was a large but not statistically significant difference in the duration of NDO between the response groups. Yoshimura et al.[21,22] showed that, with time, bladders with NDO have a decrease in the relative proportion of capsaicin-sensitive fibres and an increased proportion of neurofilament-containing fibres, which are capsaicin-insensitive. Thus, a time-dependent decrease in the sensitivity of bladder afferents to resiniferatoxin could account for the absence of response in those not responding. Larger scale studies are needed to identify the correlation between disease duration and response to resiniferatoxin treatment.

Histopathological aspects

All the specimens were of sufficient depth to assess the suburothelium (lamina propria) and in half the biopsies muscularis propria was present, which compares favourably with our previous study [19]. Szallasi et al.[23] showed in animal studies that vanilloid receptor density is 1.7 times more at the bladder neck than in the dome. As bladder neck biopsies are painful that site was avoided and specimens were instead taken from a consistent site, just above and lateral to the ureteric orifices. UTI was present at the time of baseline biopsy in five patients, with severe inflammation in two specimens, moderate in two and mild in one. All patients were treated with antibiotics, had repeat baseline assessments and were randomized to receive resiniferatoxin (two) or placebo (three patients). We included these baseline biopsies in the study, as nerve density did not change significantly in any of the specimens after treating the UTI.

There was no evidence that intravesical resiniferatoxin caused inflammation or urothelial dysplasia. Histopathological signs of inflammation, mainly chronic, were present in most biopsies taken before treatment from patients with NDO, and were present in several control biopsies. However, inflammatory changes as graded with a random score appeared to be more severe in the patients with NDO than in controls (P = 0.02). In rats, peripheral inflammation increases TRPV1 protein levels in dorsal root ganglion cells, which is then transported to peripheral C fibre terminals [24]. A similar association between inflammation and TRPV1 expression exists in the inflamed human bowel [25]. Thus, the differences noted in inflammatory changes could account for the increase in TRPV1-immunoreactive nerve fibres in patients with NDO compared with controls. However, the presence of inflammation did not appear to be related to treatment outcome, as the per-group analysis of inflammation scores from baseline biopsies showed no differences in mean values between the response groups or responders and controls, and the mean inflammation scores did not change after resiniferatoxin in either response group (P = 0.69 and 0.19).

Immunohistochemistry

Studies in the chronic spinal rat showed that capsaicin-sensitive unmyelinated afferent C-fibres originating in the suburothelium have somal hypertrophy and increased excitability, which may be related to the emergence of an aberrant C-fibre driven sacral spinal reflex [26]. Yoshimura et al.[21] showed that most (60%) bladder afferent neurones are capsaicin-sensitive and nearly all (95%) of these are C-fibre neurones. That more suburothelial nerve fibres stained for PGP9.5 in patients with spinal NDO than in normal controls implies a higher suburothelial nerve density in these patients, complementing the results of previous studies [18].

We previously reported that there were fewer TRPV1-immunoreactive nerve fibres than those immunostained for PGP9.5 in tissues obtained from postmortem normal bladders and cystoscopic biopsies from patients with NDO, suggesting that TRPV1 is expressed by a subset of neurones in the human urinary bladder [10]. In the present study we showed for the first time that suburothelial TRPV1-immunoreactivity is greater in patients with spinal NDO than in normal controls. The statistically significant correlation between baseline PGP9.5 and TRPV1 values in patients with NDO, compared to control subjects, implies that in NDO increased suburothelial nerve density is, at least partly, caused by greater vanilloid-sensitive innervation.

The higher suburothelial nerve density before treatment in patients with NDO than in controls might be related to the release of neurotrophic factors. Nerve growth factor (NGF) has been reported to increase TRPV1 mRNA expression in cultures of dorsal root ganglion neurones from adult rats [27] and to increase the number of sensory fibres and somal hypertrophy of afferent neurones in trigeminal ganglia when overexpressed [28]. The levels of NGF are greater in the urinary bladders of chronic spinalized rats than in controls [29,30]. Immunoneutralization of NGF in the spinal cord of these rats results in suppression of NGF levels in the L6-S1 dorsal root ganglion, which contains bladder afferent neurones, followed by a reduction of the degree of bladder overactivity [30]. As NGF levels are also increased in inflamed bladders [31], a possible role for NGF in the overexpression of TRPV1 in patients with NDO cannot be excluded. In addition, Steers et al.[32,33] showed that in partially obstructed rats, afferent nerves have structural and functional plasticity as a result of ongoing neurotrophic interactions associated with bladder hypertrophy. Differences in the degree of DSD-related bladder obstruction might therefore account for the variation in sensitivity to vanilloid therapy. However, there was no difference in baseline PGP9.5 and TRPV1 immunoreactivity between the response groups, which would be expected if the presence of DSD were related to treatment response. In addition, whilst it is not recommended to evaluate DSD by pressure-flow analysis [34], there appeared to be no significant difference in the incidence of DSD between the groups, based on an analysis of cystometric traces. We suspected DSD in patients who had a low urinary flow rate despite grossly elevated detrusor pressures, or in those who had a persistent inability to void in the presence of a sustained involuntary detrusor contraction.

The significant decrease of both PGP9.5 and TRPV1-immunoreactive suburothelial nerve fibres in clinical responders to resiniferatoxin provides evidence that a greater density of suburothelial afferent innervation is related to the symptoms of an overactive detrusor in patients with spinal cord lesions, but the precise mechanism of action of resiniferatoxin is unclear. Previous, controlled, animal studies have shown that animals treated with intravesical resiniferatoxin have markedly fewer bladder sensory fibres, which lie in the mucosa or muscular layer and are immunoreactive to TRPV1, substance P and CGRP [35]. These changes are transient and the number of nerve fibres staining for TRPV1, substance P and CGRP return to normal values at 1–2 months after exposure to resiniferatoxin. Resiniferatoxin is known to decrease NGF in sensory fibres of dorsal root ganglion cells and this lack of NGF might contribute to the decrease in TRPV1-immunoreactivity that follows intravesical vanilloid administration [36]. The current study is the first to show a significant reduction of TRPV1-immunoreactive fibres in patients with NDO who responded to intravesical resiniferatoxin, suggesting that C-fibre desensitization and degeneration may explain the action of intravesical resiniferatoxin in humans, as in animals. Moreover, in two of these responders the maximum clinical response after instillation of 1 µmol/L was followed by clinical relapse at 3 and 6 months, respectively. Similarly, at the immunohistochemical level, a decrease of suburothelial nerve (PGP9.5-immunoreactive) density in biopsies taken from these two patients at the time of maximum clinical response was followed by a recovery of nerve density in biopsies taken at clinical relapse, suggesting a possible connection between suburothelial plasticity and response to resiniferatoxin in humans as well as animals. However, there was no similar recovery in TRPV1 values in biopsies taken at relapse. The explanations for this inconsistency include relatively lower TRPV1 expression in regenerated fibres, or sprouting of fibres not expressing TRPV1.

Recent studies showed that TRPV1 is also expressed by human urothelial and detrusor smooth muscle cells, as well as interstitial cells [12,37,38]. Furthermore, the application of capsaicin to cultured rat urothelial cells evoked a significant release of ATP that could be blocked by the co-administration of the TRPV1 antagonist, capsazepine, together with diminished hypo-osmolality-evoked ATP release from cultured TRPV1–/– urothelial cells, suggest a role for the urothelium in bladder mechanosensation. Pharmacological modulation of Aδ-fibres is another possible mechanism of action of resiniferatoxin, and we previously reported that there was TRPV1-immunoreactivity in some thicker fibres, which might be Αδ-fibres, present both in the suburothelium and traversing the detrusor muscle [10]. Clearly, further studies with more patients are needed to determine factors at an immunohistochemical level (TRPV1 or others) whose changes after intravesical resiniferatoxin treatment could correlate to clinical response.

In conclusion, patients with spinal NDO have a greater suburothelial nerve density and vanilloid-sensitive suburothelial innervation, as detected by increased PGP9.5- and TRPV1-immunoreactivity, respectively, than normal controls. Intravesical resiniferatoxin resulted in a marked decrease of both PGP9.5- and TRPV1-immunoreactive nerve fibres in patients who responded to treatment, to values found in control tissues. Moreover, changes after treatment in TRPV1 correlated well with those for PGP9.5 in all patients. Parallel changes in bladder suburothelial TRPV1- and PGP9.5-immunoreactivity in patients with NDO after intravesical resiniferatoxin suggest that the greater number of nerve fibres in these patients are mainly of sensory origin and express TRPV1. At baseline the nerve fibre densities were similar in responders and nonresponders; an additional factor may account for the differences in response to treatment.

CONFLICT OF INTEREST

None declared. Source of funding: Afferon Corporation, Napp Laboratories, European Urological Scholarship Program and St Peter's Trust for Kidney, Bladder and Prostate Research.

Ancillary