Abdominal Pain and the Neurotrophic System in Ulcerative Colitis

Abstract

Background:

We undertook a study to test the hypothesis that inflammation alters peripheral sensory mechanisms, thereby contributing to chronic abdominal pain in ulcerative colitis (UC).

Methods:

Patients with UC and healthy individuals rated abdominal pain using a visual analog scale and completed surveys describing anxiety or depression (Hospital Anxiety and Depression Score) and gastrointestinal symptoms (Rome III questionnaire). Patient age, sex, and severity of inflammation were determined. Rectal biopsies were processed using immunohistochemical techniques to assess nerve fiber density and real-time PCR to determine transcript expression of neurotrophins (nerve growth factor, glial cell-derived neurotrophic factor, artemin, neurturin), ion channels (transient receptor potential vanilloid type 1, transient receptor potential ankyrin 1) and inflammatory mediators (tumor necrosis factor-α, interleukin [IL]-1β, IL-6, IL-10, IL-17).

Results:

A total of 77 patients with UC (27 female, 50 male) and 21 controls (10 female, 11 male) were enrolled. Patients with UC with pain had significantly higher depression scores than controls and patients with UC without pain (P < 0.05). There was no correlation between any of the inflammatory markers and pain scores. Visual analog scale pain scores significantly correlated with younger age, higher depression scores, increased expression of neurturin and decreased expression of transient receptor potential ankyrin 1 in the mucosa. Mucosal nerve fiber density did not correlate with any measures of inflammation or pain. Only higher depression scores independently predicted pain in UC (r > 0.5).

Conclusions:

We did not observe changes in mucosal innervation and did not see a significant relationship between nerve fiber density, inflammatory mediators, neurotrophic factors, or mucosal ion channel expression and pain. In contrast, the importance of depression as the only independent predictor of pain ratings mirrors functional disorders, where central processes significantly contribute to symptom development and/or perpetuation.

Pain is one of the major reasons for individuals with inflammatory bowel disease (IBD) to seek medical attention. This symptom is described by up to 70% of patients experiencing the initial onset or an exacerbation of the disease.¹ The inflammation characterizing IBD is considered to be the primary driver of pain because cytokines and other inflammatory mediators may sensitize extrinsic and intrinsic primary afferent neurons found in the gut. However, inflammation by itself does not explain pain in all individuals with IBD. Previous estimates suggest that up to 50% of individuals with ulcerative colitis (UC) or Crohn’s disease in clinical remission experience persistent symptoms, including chronic pain.^2–4 Some evidence suggests that many of these patients exhibit low-grade inflammation or, in the case of Crohn’s disease, may have developed strictures with intermittent obstructive symptoms.⁵ Even when this is taken into account, although, many patients with quiescent IBD have chronic abdominal complaints that cannot be explained by their underlying disease, raising questions about an overlap between IBD and disorders of visceral hypersensitivity, such as irritable bowel syndrome (IBS).⁶ Factors contributing to the pathogenesis of symptoms in IBS, such as anxiety, depression, and somatoform disorders, also play a role in the clinical manifestations of IBD, providing additional support for such an overlap.^7–9 Alternatively, recent studies investigating the underlying mechanisms of IBS suggest that inflammation has a role in the pathogenesis of this functional disorder. For example, the disease manifests after a gastrointestinal infection in a subgroup of individuals, with clinical indicators of a more significant acute inflammation notably contributing to the risk of subsequent IBS development.¹⁰ Additionally, several investigators have described subclinical inflammation within the mucosa of IBS patients.^11–13 Such findings are consistent with a conceptual model that includes peripheral sensitization, triggered and/or driven by colonic inflammation, as a mechanism of persistent abdominal pain.

The peripheral and central nervous systems both exhibit significant functional and structural plasticity, which contributes to changes in sensory input and processing after inflammation. For example, colitis is associated with enhanced excitability of primary afferent neurons.^14,15 This is generated by inflammatory mediators and neurotrophic factors, which are upregulated during inflammation and alter the expression and/or properties of ion channels or other molecules. These changes provide the molecular basis for altered response characteristics of afferent neurons (“peripheral sensitization”), which, in turn, can modulate the properties of spinal and higher order neurons (“central sensitization”), ultimately resulting in long-lasting visceral hypersensitivity. The majority of visceral afferents send unmyelinated fibers to their peripheral targets. Most of these neurons express receptors for nerve growth factor and artemin and respond to agonists for the transient receptor potential vanilloid type 1 (TRPV1) and the transient receptor potential ankyrin 1 (TRPA1) channels.^16,17 Animal models of colitis have been associated with an increase in nerve growth factor expression, enhanced excitability of primary afferents, and augmented responses to colorectal distension.^18,19 Interestingly, genetic deletion of the TRPV1 channel blocks the development of hypersensitivity.²⁰ An important role of TRPV1 and its modulation in the pathogenesis of visceral pain was also suggested by several investigations reporting increased TRPV1 nerve fiber density within the colorectal mucosa in patients with rectal urgency, active colitis, and pain despite quiescent or improved colitis.^21–23 Thus, several complementary lines of evidence support a role of inflammation as trigger for visceral hypersensitivity that may also drive central sensitization and persist even during periods of low or no inflammatory activity. Conditions with chronic symptoms, especially when present during apparent clinical remission, may therefore provide the opportunity to examine the relative importance of this potential disease mechanism. We focused on UC as a disease with chronic primarily mucosal inflammation that is not complicated by stricture formation. Our underlying hypothesis was that mucosal markers of peripheral sensitization play an important role in the development of chronic pain in quiescent IBD. To test this hypothesis, we compared healthy controls with a large cohort of patients with UC with different levels of disease activity and abdominal pain.

MATERIALS AND METHODS

Patient Recruitment

Outpatients presenting for an elective colonoscopy at the University of Pittsburgh Medical Center were approached about the study. To be included in this study, participants had to be between 18 and 70 years of age, able to understand, speak and write English, and able to make their own health care decisions. Written informed consent that was HIPAA compliant and approved by the University of Pittsburgh IRB (protocol no. 11070169) was obtained from all participants. Clinical data were then obtained to assign participants to one of the following 5 subgroups: (1) controls, (2) quiescent UC without pain, (3) active UC without pain, (4) quiescent UC with pain, and (5) active UC with pain (UCAP). Participants enrolled as controls had to be pain-free and could not have a history of UC or Crohn’s disease, microscopic colitis, celiac disease, functional bowel disorder, a prior colectomy, active cancer, current pregnancy, a recent or current intra-abdominal abscess, or infection. In addition, they could not have received treatment with immunomodulators, opioids, or anticonvulsive agents within 6 months before inclusion into this study. Colonoscopy reports generated at the day of the encounter were reviewed to confirm normal findings without inflammation or malignancy. For the patient cohort, the diagnosis of UC was based on the following standard diagnostic criteria: (1) diarrhea, rectal bleeding, abdominal pain, fever, complicated perianal disease, extraintestinal manifestations, weight loss, or failure to thrive on 2 or more occasions separated by at least 8 weeks and/or ongoing symptoms of at least 6 weeks duration and (2) ulcerations, pseudopolyps endoscopic, and/or histological evaluation. When there were only minor changes (mucosal edema, erythema, loss of normal submucosal vascular patterns, friability), mucosal biopsies had to be consistent with the presence of UC, confirmed by a trained pathologist and based on mucosal erosions and/or ulcerations, architectural changes of crypts, and/or Paneth cell metaplasia. Exclusion criteria for all participants were the presence of a prior colectomy, an intra-abdominal abscess, toxic megacolon, active cancer, evidence of a current infection or other inflammatory processes unrelated to UC, or the diagnosis of any other form of colitis. The electronic medical record, the colonoscopy findings, and histological report were reviewed to confirm the correct diagnosis and assess disease activity.

Data Extraction

After enrollment, we obtained demographic data for all participants. For individuals with UC, we also reviewed the electronic medical record and abstracted disease onset, extent, severity of inflammation and complications, previous and current treatment(s), and comorbid conditions (including psychiatric disease), as well as current medication use (including pain medications, such as non-steroidal anti-inflammatory drugs and opiates). Values for C-reactive protein and erythrocyte sedimentation rate were also obtained from the medical record.

Endoscopic and Histologic Grading

Colonoscopic findings obtained on the day of the study were coded based on the extent and severity of inflammation and the presence of potential disease-related complications or incidental findings. All colonoscopies were performed by experienced gastroenterologists specializing in IBD management. Disease activity was graded as absent (0), mild (1), moderate (2), or severe (3) using the Mayo Endoscopic Scoring of UC system.²⁴ Final assessment of disease activity was determined by histological evaluation of the biopsy samples. Histologic scoring was performed by an expert gastrointestinal pathologist. Samples with any degree of inflammation (mild–severe) on histological evaluation were defined as having “active” disease.

Survey Data

We determined whether individuals were experiencing abdominal pain by their response to the following question: “Have you been experiencing pain or discomfort in your abdomen?” The duration of any reported abdominal pain was subsequently determined. Severity of the abdominal pain was then defined using a visual analog scale (VAS) and present pain intensity score, whereas the quality of the abdominal pain was determined from sensory and affective subscores (all derived from the McGill Short Form Questionnaire [SF-MPQ]²⁵). Determination of IBS diagnosis was based on responses to questions derived from the Rome III Study Group’s criteria for IBS (www.romecriteria.org). Study participants also provided responses to surveys based on the Hospital Anxiety and Depression Score²⁶ to provide qualitative information about mood-related symptoms study participants were experiencing around the time of their endoscopy. Finally, quality of life scores were determined based on responses provided by each patient to the Short Inflammatory Bowel Disease Questionnaire (SIBDQ)²⁷ obtained during their most recent office visit to the UPMC IBD Center.

Tissue Acquisition

Two colorectal mucosal biopsies were taken in succession using standard capacity biopsy forceps approximately 10 cm proximal to the anal verge. One was immediately placed in a fixative agent (4% paraformaldehyde), whereas the other was placed in RNAlater (Invitrogen, Carlsbad, CA) solution. Samples designated for RNA extraction were frozen (−80°C) until time of RNA isolation.

Immunohistochemistry

Biopsies dedicated for immunohistochemical evaluation were placed in 4% paraformaldehyde for 4 hours, incubated overnight in 25% sucrose solution then set en bloc and laid out on slides to be evaluated. These slides were subsequently stained with antibodies specific to human anti-PGP9.5, a member of the ubiquitin hydrolase family found in neurons and their processes, using protocols described previously by our laboratories.^28,29 Tissue sections were digitized at a single exposure with a 20× objective, and the intensity of immunoreactivity was determined using the NIH IMAGE morphometrics program (http://rsb.info.nih.gov/nih-image/).

mRNA Expression

Additional samples were collected in RNAlater (Invitrogen) and frozen until the time of RNA isolation. Biopsies were processed using RNeasy isolation columns (Qiagen, Hilden, Germany) and real-time quantitative PCR was undertaken as described by Malin et al.²⁹ Primer pairs (Integrated DNA Technologies, Coralville, IA) for selected neurotrophic factors, TRP channels, and cytokines (Table 1) were compared with a “house-keeping” gene (β-actin) to calculate relative fold changes in the expression of various transcripts.³⁰

TABLE 1.

Primer Sequences for Selected Cytokine, Neurotrophin, and Ion Channel Targets

Target	Sense Primer	Antisense Primer
β-actin (h)	GGCCGAGGACTTTGATTGCATTGT	AGGATGGCAAGGGACTTCCTGTAA
TNFα	ATCAAGAGCCCCTGCCAGAG	AAAGTAGACCTGCCCAGACTCG
IL-1β	ATGATGGCTTATTACAGTGGCAA	GTCGGAGATTCGTAGCTGGA
IL-6	TTCGGTCCAGTTGCCTTCTCCC	AGGTGAGTGGCTGTCTGTGTGG
IL-10	TCCTTGCTGGAGGACTTTAAGGGT	TGTCTGGGTCTTGGTTCTCAGCTT
IL-17	AGGCCATAGTGAAGGCAGGAATCA	ATTCCAAGGTGAGGTGGATCGGTT
NGF	TTCACCCCGTGTGCTGTTTAG	ATGATGACCGCTTGCTCCTGTG
GDNF	AAGAGAGCGGAATCGGCAGG	CATAGCCCAGACCCAAGTCAGTG
Artemin	CTCCACACGACCTCAGCCTG	CGGTTCTCCAGGTGCTGTTGAC
Neurturin	AGCTCCGTGCTGTCCATCTGGAT	TGCAGGAGTGCACGGTACT
TRPV1	TGAAGCCGTTGCTCAGAATAACTG	CTCAGGGTCTTTGAACTCGTTGTC
TRPA1	TGTTTCTCAGTGACCACAATGGC	AGTGTTCCCGTCTTCATCCAGG

GDNF, glial cell-derived neurotrophic factor; IL, interleukin; NGF, nerve growth factor.

Statistical Analysis

Unless indicated specifically, all data are presented as mean ± SEM. Statistical analyses were performed using Prism software (version 4.0a; GraphPad Software, San Diego, CA). Group assignments were based on the apriori definition of subgroups. Comparisons between 2 groups were performed using the unpaired Student’s t test or χ² test. The primary endpoint was abdominal pain rating, as defined by the VAS of the McGill questionnaire. Patients were specifically instructed to focus on their abdominal pain when completing the questionnaire. Univariate analyses were performed to determine correlations between different variables and pain ratings. Variables with a correlation coefficient of P < 0.1 were entered into a stepwise multilinear regression analysis to identify independent predictors of pain severity. Comparisons among 3 or more groups were performed using one-way analysis of variance and Bonferroni’s posttest. Differences between means at a level of P < 0.05 were considered to be significant.

Ethical Considerations

This study was reviewed and approved by the Institutional Review Board of the University of Pittsburgh (IRB Protocol # PRO11070169).

RESULTS

Study Sample

We recruited a total of 109 individuals. Eleven were excluded from the study (9 were found to have a diagnosis other than UC, 1 had Clostridium difficile colitis, and another had colon cancer at the time of the study). Of the 98 remaining participants (37 female, 61 male), 21 were described as healthy controls (10 female, 11 male), whereas 77 had UC (27 female, 50 males) (Fig. 1). Controls underwent colonoscopy for colorectal cancer screening (n = 17) or assessment of painless hematochezia (n = 4) without associated changes in bowel pattern. All controls had a macroscopically normal colonoscopy without evidence of inflammation or malignancy. Individuals with UC underwent colonoscopy for the purpose of restaging their disease status (n = 36), colon cancer surveillance (n = 27), or both (n = 12).

FIGURE 1.

Patient recruitment and exclusion diagram.

Based on our apriori design, we separated individuals with UC into 4 distinct groups based on their histological disease activity status and patient reports of abdominal pain: (1) active UC with pain (24 total; 7 female, 17 males), (2) active UC without pain (21 total; 3 females, 18 males), (3) inactive UC with pain (UCIP) (15 total; 8 females, 7 males), and (4) inactive UC without pain (17 total; 9 females, 8 males). We also separately stratified individuals with UC based on presence of abdominal pain (UC with pain: 39 total; 15 females, 24 males) or absence of pain (UC without pain: 38 total; 12 females, 26 males) and by presence of inflammation (active UC: 45 total; 10 females, 35 males) or absence of inflammation (inactive UC: 32 total; 17 females, 15 males).

Healthy controls (52.2 ± 1.6 yr) were significantly older than the combined UC group (45.0 ± 1.7 yr) (Table 2). The gender distribution did not differ significantly from controls. When the UC patient sample was dichotomized based on the presence or absence of pain, the subgroup with pain was significantly younger (40.6 ± 2.4 yr; n = 39) than that without pain (49.5 ± 2.0 yr; n = 38) (Table 3). There was no difference in the gender distribution of patients with and without pain (Table 3). There were no significant differences in the prevalence of mood disorders or in the use of antidepressive or anxiolytic medications when controls were compared with the whole UC cohort or when compared with the various UC subgroups (Tables 2–4).

TABLE 2.

Comparison of Healthy Controls, Total UC Populations, and UC Subpopulations Categorized by Disease Activity and Abdominal Pain Status

Variable	Controls (n = 21)	All UC (n = 77)	UCAP (n = 24)	UCANP (n = 21)	UCIP (n = 15)	UCINP (n = 17)	p
Age, yr	52.2 ± 1.6	45.0 ± 1.7	40.2 ± 3.2	48.1 ± 2.9	41.2 ± 4.8	51.4 ± 2.8	<0.05a
Gender (female%/male%)	10 (48.0)/11 (52.0)	27 (35.1)/50 (64.9)	7 (29.2)/17 (70.8)	3 (14.3)/18 (85.7)	8 (53.3)/7 (46.7)	9 (52.9)/8 (47.1)	NS
Disease duration, yr	NA	10.3 ± 1.1	11.3 ± 2.1	8.7 ± 1.3	6.5 ± 1.6	13.5 ± 3.1	NS
CRP, mg/dL	NA	0.7 ± 0.2	1.4 ± 0.5	0.4 ± 0.2	0.4 ± 0.1	0.3 ± 0.1	<0.05b
ESR, mm/h	NA	17.8 ± 2.3	24.8 ± 4.5	8.4 ± 2.6	10.2 ± 1.4	18.6 ± 4.0	<0.05b
IL-1β mRNA	1.2	11.8	16.2	13.5	12.0	3.2	<0.05a
TNFα mRNA	1.4	1.7	2.2	2.1	1.4	1.0	NS
IL-6 mRNA	5.4	5.0	5.7	8.1	2.1	2.6	NS
IL-10 mRNA	1.2	1.9	3.2	1.8	1.2	1.0	<0.05a
IL-17 mRNA	1.6	5.0	8.0	5.3	4.1	1.6	<0.05a
Coexisting diagnosis of affective spectrum disorder (%)	4 (19.0)	15 (19.5)	5 (20.8)	1 (4.7)	5 (33.3)	4 (23.5)	NS
Use of anxiolytic or antidepressant (%)	1 (4.8)	15 (19.5)	4 (16.7)	1 (4.7)	5 (33.3)	5 (29.4)	NS
HADS anxiety score	5.1 ± 0.7	4.5 ± 0.3	5.3 ± 0.5	3.4 ± 0.6	4.9 ± 0.7	4.4 ± 0.7	NS
HADS depression score	2.3 ± 0.6	3.0 ± 0.3	3.6 ± 0.5	1.9 ± 0.4	3.9 ± 0.8	2.7 ± 0.6	NS
Use of analgesic (NSAID/opiate/both/either) (%)	2 (9.6)/0 (0)/0 (0)/2 (9.6)	13 (16.9)/4 (5.2)/0 (0)/17 (22.1)	2 (8.3)/1 (4.2)/0 (0)/3 (12.5)	4 (19)/0 (0)/0 (0)/4 (19)	3 (20)/1 (6.7)/0 (0)/4 (26.7)	2 (11.8)/2 (11.8)/0 (0)/4 (23.6)	NS
NGF mRNA	1.1	1.1	1.1	1.4	0.9	1.0	NS
GDNF mRNA	1.2	0.9	0.9	0.9	0.8	0.9	NS
Artemin mRNA	1.3	1.7	1.3	2.9	1.6	1.0	NS
Neurturin mRNA	1.2	0.6	0.8	0.6	1.0	0.3	<0.05a
TRPV1 mRNA	1.1	0.8	1.1	0.7	0.6	0.7	<0.05a
TRPA1 mRNA	1.2	1.2	1.0	1.7	0.8	1.0	NS
Nerve fiber density (lamina propria%)	27.4 ± 2.6	27.3 ± 1.8	25.8 ± 4.1	28.5 ± 2.6	25.5 ± 3.6	29.2 ± 4.2	NS

P values listed were determined by t test compared with controls unless otherwise indicated.

^aOne-way analysis of variance with Bonferroni’s posttest compared with controls.

^bOne-way analysis of variance and Bonferroni’s posttest compared with UCAP.

CRP, C-reactive protein; DNF, glial cell-derived neurotrophic factor; ESR, erythrocyte sedimentation rate; HADS, Hospital Anxiety and Depression Score; NGF, nerve growth factor; NSAID, non-steroidal anti-inflammatory drug; UCANP, Active UC without Pain, UCINP, Inactive UC without Pain.

TABLE 3.

Comparison of Healthy Controls and Patients with UC with and Without Abdominal Pain

Variable	Controls (n = 21)	UCP (n = 39)	UCNP (n = 38)	P
Age, yr	52.2 ± 1.6	40.6 ± 2.4	49.5 ± 2.0	<0.05
Gender (female/male) (%)	10 (48.0)/11 (52.0)	15 (38.5)/24 (61.5)	12 (31.6)/26 (68.4)	NS
Disease duration, yr	NA	9.5 ± 1.5	11.1 ± 1.7	NS
CRP, mg/dL	NA	1.0 ± 0.3	0.4 ± 0.1	<0.05a
ESR, mm/h	NA	19.5 ± 3.3	15.4 ± 2.8	NS
IL-1β mRNA	1.2	14.5	8.9	<0.05
TNFα mRNA	1.4	1.9	1.6	NS
IL-6 mRNA	5.4	4.3	5.7	NS
IL-10 mRNA	1.2	2.4	1.4	NS
IL-17 mRNA	1.6	6.5	3.7	<0.05
Coexisting diagnosis of affective spectrum disorder (%)	4 (19.0)	10 (25.6)	5 (13.2)	NS
Use of anxiolytic or antidepressant (%)	1 (4.8)	9 (23.1)	6 (15.8)	NS
HADS anxiety score	5.1 ± 0.7	5.1 ± 0.4	3.8 ± 0.4	NS
HADS depression score	2.3 ± 0.6	3.7 ± 0.5	2.3 ± 0.3	<0.05
Use of analgesic (NSAID/Opiate/Both/Either) (%)	2 (9.6)/0 (0)/0 (0)/2 (9.6)	5 (12.8)/2 (5.1)/0 (0)/7 (17.9)	6 (15.8)/2 (5.3)/0 (0)/8 (21.1)	NS
NGF mRNA	1.1	1.0	1.2	NS
GDNF mRNA	1.2	0.9	0.9	NS
Artemin mRNA	1.3	1.4	2.0	NS
Neurturin mRNA	1.2	0.9	0.4	<0.05
TRPV1 mRNA	1.1	0.6	0.7	<0.05
TRPA1 mRNA	1.2	0.9	1.4	NS
Nerve fiber density (lamina propria, %)	27.4 ± 2.6	25.7 ± 2.8	28.8 ± 2.3	NS

P values listed were determined by one-way analysis of variance with Bonferroni’s posttest compared with controls unless otherwise indicated.

^aP < 0.05 determined by student’s t test compared with UCP.

CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; GDNF, glial cell-derived neurotrophic factor; HADS, Hospital Anxiety and Depression Score; NGF, nerve growth factor; NSAID, non-steroidal anti-inflammatory drug; UCNP, UC without pain; UCP, UC with pain.

TABLE 4.

Comparison of Healthy Controls and Patients with UC with and Without Inflammation on Histological Evaluation

Variable	Controls (n = 21)	UCA (n = 45)	UCI (n = 32)	P
Age, yr	52.2 ± 1.6	43.8 ± 2.2	46.6 ± 2.5	<0.05
Gender (female/male) (%)	10 (48.0)/11 (52.0)	10 (22.2)/35 (77.8)	17 (53.1)/15 (46.9)	NS
Disease duration, yr	NA	10.2 ± 1.3	10.4 ± 1.9	NS
CRP, mg/dL	NA	1.1 ± 0.3	0.3 ± 0.1	<0.05a
ESR, mm/h	NA	20.7 ± 3.6	14.4 ± 2.5	NS
IL-1β mRNA	1.2	15.0	7.3	<0.01
TNFα mRNA	1.4	2.2	1.2	<0.05
IL-6 mRNA	5.4	6.8	2.4	NS
IL-10 mRNA	1.2	2.5	1.1	<0.05
IL-17 mRNA	1.6	6.7	2.8	<0.05
Coexisting diagnosis of affective spectrum disorder (%)	4 (19.0)	6 (13.3)	9 (28.1)	NS
Use of anxiolytic or antidepressant (%)	1 (4.8)	5 (11.1)	10 (31.3)	<0.05
HADS anxiety score	5.1 ± 0.7	4.4 ± 0.4	4.6 ± 0.5	NS
HADS depression score	2.3 ± 0.6	2.8 ± 0.4	3.3 ± 0.5	NS
Use of analgesic (NSAID/Opiate/Both/Either) (%)	2 (9.6)/0 (0)/0 (0)/2 (9.6)	6 (13.3)/1 (2.2)/0 (0)/7 (15.5)	5 (15.6)/3 (9.4)/0 (0)/8 (25)	NS
NGF mRNA	1.1	1.2	0.9	NS
GDNF mRNA	1.2	0.9	0.8	NS
Artemin mRNA	1.3	2.1	1.3	NS
Neurturin mRNA	1.2	0.7	0.6	<0.05
TRPV1 mRNA	1.1	0.9	0.7	<0.05
TRPA1 mRNA	1.2	1.3	0.9	NS
Nerve fiber density (lamina propria%)	27.4 ± 2.6	27.1 ± 2.5	27.5 ± 2.8	NS

P values listed were determined by one-way analysis of variance with Bonferroni’s posttest compared with controls unless otherwise indicated.

^aP < 0.05 determined by Student’s t test compared with UCI.

CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; GDNF, glial cell-derived neurotrophic factor; HADS, Hospital Anxiety and Depression Score; NGF, nerve growth factor; NSAID, non-steroidal anti-inflammatory drug; TNFα, tumor necrosis factor-α; UCA, Active UC, UCI, Inactive UC.

Disease Duration, Inflammatory Activity, and UC-directed Therapy

The mean duration of disease for the whole UC population was 10.3 ± 1.1 years. Based on the initial endoscopic assessment of each patient, disease extent was pancolonic in 56 individuals (72.7%) and was limited to the left side of the colon in 15 patients (19.5%) and the rectum in the 6 remaining patients (7.8%). Of the 77 patients with UC enrolled in this study, 70 (90.9%) were receiving treatment with 5-aminosalicylic acid, immunomodulator and/or biologic therapy (58 individuals were taking 1 or more 5-aminosalicylic acid product, 28 individuals were on immunomodulator therapy [6-MP, azathioprine or methotrexate], and 22 individuals were on biologic therapy [infliximab or adalimumab]). Twenty-eight individuals (36.4%) were taking more than one of these agents together including 13 patients (16.9%) on combined immunomodulator and biologic therapy.

Pain Experience and Analgesic Use

Each patient completed questionnaires to provide information about their experiences with abdominal pain and associated symptoms. The various pain intensity scores correlated well with one another (e.g., McGill sensory versus affective pain rating: r = 0.7; McGill sensory pain rating versus McGill VAS: r = 0.67; SIBDQ pain rating versus McGill VAS: r = −0.50). Most (82.5%) of the affected individuals described their pain as mild to moderate in intensity on one or more of the abdominal pain intensity rating scales. A pairwise comparison between subgroups based on mucosal inflammation did not show a significant impact on pain rating (VAS 7.4 ± 1.6 versus 5.0 ± 2.0; P = 0.14). Of the UC subgroups reporting abdominal pain, the UCAP group had a mean McGill VAS score of 12.9 ± 2.5 mm, whereas the UCIP group had a mean score of 9.1 ± 2.6 mm (not significant). The UCAP group had a mean McGill Present Pain Intensity Score of 1.0 ± 0.2, whereas the UCIP group had a present pain intensity score of 1.1 ± 0.3 (not significant). Of note, these 2 groups also had statistically similar McGill Sensory Pain (8.5 ± 1.1 and 7.7 ± 1.9, respectively) and McGill Affective Pain (2.0 ± 0.4 and 1.9 ± 0.6, respectively) scores. We also found, using Rome III guidelines, that 22 individuals with UC (28.6%) met criteria for IBS. Of note, the majority of individuals with UC who described having pain met these criteria (56.4%).

Although the study was not powered to address the potential role of analgesic agents, we examined the use of non-steroidal anti-inflammatory drugs and opioids. Only 16 study participants with UC used either of these agents at the time of this study (4 used opioids and 12 used non-steroidal anti-inflammatory drugs). There was no difference between controls and patients or among the predefined UC subpopulations regarding use of analgesics (Tables 2–4).

Inflammatory Activity

Thirty-two patients with 32 UC (41.5%) had quiescent disease based on histologic assessment. The remaining participants in the UC group had mild (n = 22, 28.6%) or moderate (n = 16, 20.8%) inflammation on biopsy, with only 7 (9.1%) presenting with severe colitis. Histologic and endoscopic grading were significantly correlated with one another (r = 0.86). Biomarkers of inflammation (C-reactive protein concentrations and transcript levels of tumor necrosis factor-α, interleukin-1β, interleukin-10, and interleukin-17) were all significantly elevated in the UC subgroup with histologic evidence of disease activity (Table 4). When we dichotomized patients based on the presence or absence of moderate or severe inflammation, age and sex distribution did not differ between these 2 subsets. As indicated above, there was also no significant difference in pain ratings between the UCAP and UCIP subgroups.

Neurotrophic Factor and TRP Channel Transcript Levels in the Mucosa

Considering the role of neurotrophic factors in neuronal plasticity, we next evaluated mucosal transcript levels of nerve growth factor, artemin, glial cell-derived neurotrophic factor, and neurturin (NRTN). Comparing results between all patients with UC and controls, there was a small but significant difference for NRTN only (Table 2). Considering the focus on pain, we again dichotomized the study participants based on the presence and absence of pain. Only NRTN transcript levels differed between the groups, with lower values demonstrated in patients without pain when compared with controls or patients with UC with pain (Table 3). Interestingly, transcript levels for TRPV1 were higher in controls compared with patients with UC (Table 3). When examining the potential correlation with pain in patients, we noted lower levels for TRPV1 and higher levels for TRPA1 in patients without pain (Table 3).

Nerve Fiber Density in the Mucosa

The pan-neuronal marker PGP9.5 provided robust staining of nerve fibers, which extended throughout the lamina propria toward the epithelial layer (Fig. 2). When comparing all patients with UC with controls or UC subgroups stratified by the presence of pain or inflammation, we did not discern differences in mucosal nerve fiber density (Tables 2–4).

FIGURE 2.

Representative microscopic images of rectal mucosa stained for the pan-neuronal marker PGP9.5. Although there were no significant differences in nerve fiber density among the various control and UC groups, there was a large degree of variation in nerve fiber staining patterns, sometimes even within the same sample.

Anxiety and Depression Scores

Considering the importance of anxiety and depression in functional illness, we compared Hospital Anxiety and Depression Score subscores among the groups. There were no significant differences between controls and patients with UC (Table 2). However, patients with UC with pain had significantly higher Hospital Anxiety and Depression Score when compared with controls and patients with UC without pain (Table 3).

Univariate and Multivariate Analyses

Considering the multiple variables and subgroups, our design is susceptible to type I errors. We therefore focused on pain as the endpoint and performed a univariate analysis to detect significant correlations with measures obtained in the study. Consistent with the results described above, younger age, higher depression scores, higher NRTN levels, and lower TRPA1 levels correlated with higher pain ratings (see Table, Supplemental Digital Content 1, http://links.lww.com/IBD/A624). These variables were entered into a linear regression model. Only higher depression scores independently predicted pain ratings (r > 0.50).

DISCUSSION

This study provides important insight into mechanisms contributing to chronic pain in IBD. We considered UC, a model disease, to test the hypothesis that inflammation can cause sensitization of peripheral afferents projecting to the intestines and visceral hyperalgesia. Because such mechanisms have been invoked in structural and functional diseases of the gastrointestinal tract,^31,32 the results may also be relevant for diseases other than UC. In contrast to previous studies, which reported a correlation between subclinical mucosal inflammation and pain in IBS,^{11–13,33–35} pain in UC did not correlate with evidence of mucosal inflammation. We noted similarly negative findings with markers of neuronal plasticity within the colonic mucosa. Neither transcript levels of neurotrophic factors and ion channels likely contributing to nociceptive responses nor mucosal innervation were consistently altered by chronic inflammation or correlated with abdominal pain, which differs from previous reports.^21–23,36

These negative results raise questions about differences in investigational techniques, sample size, and statistical power. For example, previous studies have demonstrated significant changes in peripheral nerve density although with smaller patient cohorts.^21–23 However, these studies also reported that a more than 3-fold increase in fiber density in IBS and IBD compared with controls, arguing against sample size as being the limiting factor. Consistent with this conclusion, we were able to detect differences in inflammatory markers when comparing subgroups defined by the presence or absence of active disease and noted a correlation between pain and effect as a potentially important factor. Differences in technique may be an important contributing factor. Our immunohistochemical experiments focused on nerve fiber density, relying on PGP9.5 as a pan-neuronal marker. A previous study in IBS had demonstrated comparable increases in both TRPV1- and PGP9.5-immunoreactive nerve fibers within the mucosa of IBS patients, providing a rationale for this approach.²² These results contrast with a nearly 4-fold increase in TRPV1-positive nerve fibers despite stable PGP9.5-immunoreactivity within the mucosa of IBD patients with pain compared with controls and patients without pain.²¹ Although we have successfully used these immunohistochemical techniques in the past,^29,37 neither TRPV1 nor TRPA1 antibodies allowed specific staining in human tissues in our hands (data not shown). We thus relied on complementary approaches, looking at mRNA levels in mucosal biopsies and did not see a difference. The assay could not address the cellular sources of mRNA levels. However, previous studies have established localization of mRNA in nerve endings, which contributes to synaptic plasticity and should thus be reflected in our data.³⁸ The lack of changes in mucosal expression of neurotrophic factors regulating TRPV1 and TRPA1 similarly supports the validity of our findings. Finally, it is possible that differences in pain intensity contributed to the apparent discrepancies. We used validated pain assessment tools with internally consistent results. Patients enrolled in this investigation and previously published studies were recruited from outpatient clinics and more frequently had mild-to-moderate disease and pain severity. The majority of patients with UC with pain in our study also met consensus criteria for IBS, thus phenotypically overlapping with at least one of the previously described cohorts.²² Looking at markers of inflammation within our cohort, we noted a significant correlation between colonoscopic grading, histology, C-reactive protein, and cytokines. This internal consistency supports the validity of our approach. Overall, our findings thus argue against peripheral sensitization as an important and ongoing mechanism for pain in UC.

Although our data stand in contrast with the previous work on markers of peripheral sensitization, several elements of this study did corroborate the findings of previous investigations involving visceral pain syndromes, including those associated with IBD. Patients with UC with pain were significantly younger than controls and their counterparts without pain and were found to have significantly higher depression scores on a validated screening tool for affective spectrum disorders, which is consistent with previous findings in larger patient cohorts with UC or other gastrointestinal disorders.^4,39–41 Anxiety with catastrophizing and hypervigilance and depression with somatoform symptoms have both been linked to increased symptom severity, including pain in patients with functional disorders.^42–45 Health-related anxieties similarly affect the perceived quality of life in IBD patients and increase over time.^46,47 We did not see differences in anxiety between groups, which may be due to our design because we relied on a screening tool rather than a more disease-specific survey⁴⁸ and did not power our study to assess the impact of anxiety on disease severity.

Despite recruiting a large cohort and performing a comprehensive assessment, our study has important limitations. The predefined division in subgroups based on inflammation and pain resulted in smaller samples with limited statistical power. As discussed above, the internal consistency of data and positive findings related to previously identified disease mechanisms make it less likely that an important or common disease mechanism was missed. The control population was not perfectly matched in age. However, the difference was relatively small because the mean age was close to 50 years in both groups, and a 10% difference is thus unlikely to lead to differences in biological signaling within the colonic mucosa. We did not examine protein expression but instead relied on mRNA assays to examine changes in inflammatory markers or neurotrophic factors, leaving room for speculation about changes in protein levels as possible factors in the clinical manifestation of UC. Finally, we focused on pain as the primary outcome measure and cannot determine whether and how other clinical manifestations are affected by peripheral sensitization.

Overall, the comprehensive assessment in this large cohort of patients raises questions about the true relevance of peripheral sensitization as a mechanism of pain in UC. Instead, our findings fit more into a framework that shows a reciprocal relationship between affect and disease manifestations. With increasing disease duration, anxiety and depression become more prevalent and are associated with higher symptom burden independent of disease severity.^46,47,49,50 Considering the role of inflammation and systemic inflammatory response on emotional processing,^51–55 additional studies could explore how central sensitization in turn may have been driven by peripheral factors. This information opens opportunities for future mechanistic studies, linking peripheral inflammation, affect and symptom severity and provides a rationale for treatment strategies in affected individuals.