Fear-induced hyperalgesia in quiescent inflammatory bowel disease.

Patients with ulcerative colitis, a subtype of IBD mainly affecting the colon, and age-matched healthy volunteers were recruited through public advertisement between July 2019 and March 2020. In addition, patients were directly contacted during this period through a database maintained by the Clinic for Internal and Integrative Medicine (Kliniken Essen-Mitte, Essen, Germany). All volunteers underwent a standardized telephone screening and completed a comprehensive psychosocial and clinical symptom questionnaire battery (details below), including screening for symptoms of anxiety and depression based on a validated questionnaire (Hospital Anxiety and Depression Scale [HADS]; details below). 32 General exclusion criteria were age 60 years, body mass index (BMI) 30, and the usual magnetic resonance imaging (MRI)–specific criteria (eg, claustrophobia, ferromagnetic implants), as structural brain images were acquired in this study for use in a larger voxel-based morphometry analysis. 61 Pregnancy was ruled out using a commercially available pregnancy test (Biorepair GmbH, Sinsheim, Germany, sensitivity 10 mIU/mL) on the day of the study. For HC, additional exclusion criteria were any self-reported somatic or mental health conditions or regular use of medication (except for hormonal contraception or hormone replacement therapy). For IBD patients, who required an existing and confirmed diagnosis before recruitment for this study, diagnosed psychiatric comorbidities as assessed by self-report led to exclusion, whereas minor psychological symptoms, such as mild anxiety or depression symptoms indicated by elevated HADS scores were not exclusionary. To minimize putative effects of medical treatments required during phases of disease exacerbation on study measures, treatment with immunosuppressants within the past 3 months and treatment with corticosteroids or antidepressants within the past month was exclusionary. Please note that, in contrast to our previous study on pain-related fear learning and memory in IBD, 53 patients were not excluded based on measures of disease activity. Instead, inflammatory markers were assessed in blood and stool samples together with symptom reports (details below) to monitor acute (low-grade) inflammation. Work was conducted in accordance with the Declaration of Helsinki, and the study was approved by the local Ethics Committee of the University Hospital Essen (protocol number 16-7237-BO). All volunteers gave written informed consent and received monetary compensation for participation. Gastrointestinal (GI) symptoms were quantified with an established questionnaire suitable for both visceral pain conditions and healthy controls (who also experience such symptoms, albeit less frequently or intensely). 52 Participants rated the frequency of typical GI symptoms (eg, diarrhea, constipation, nausea, abdominal pain, heartburn, bloating) over the past 3 months on a Likert-type response scale (0 = never, 3 = more than twice a week), and a total sum score was calculated for analysis. To further characterize pain symptoms and pain-related emotions and cognitions, all participants completed the Brief Pain Inventory, 69 which includes subscales for pain severity (least, average, worst, and current pain) and pain interference (eg, impact on activity, mood, social relations) in the past 24 hours. Items are rated on a 0 to 10 numeric scale and subscale scores were averaged for analysis. Although this study's experimental focus was on pain-related fear as immediate emotional response to pain-predictive cues, we used the validated Pain Anxiety Symptoms Scale (PASS-20) 46 , 58 to complementary assess pain-related anxiety, which is a more future-oriented state related to the threatening long-term consequences of pain. 83 The scale covers affective, cognitive, physiological, and motivational domains on Likert-type response scales (0 = never, 5 = always), with total scores ranging from 0 to 100. Similarly, the Pain-Related Self Statements Scale 25 assessed pain catastrophizing and coping (0 = almost never, 5 = almost always; subscale scores 0-45). Fear of pain was evaluated with the Fear of Pain Questionnaire, 59 which comprises 3 subscales (minor, severe, and medical pain). Items are scored from 1 (no fear) to 5 (extreme fear), with a total score range of 30 to 150. Moreover, given their broad role in visceral pain, 5 , 47 , 50 general anxiety and depression symptoms, and perceived chronic stress levels were also assessed. The HADS 32 comprises 2 subscales (anxiety and depression; score range per subscale: 0-21). Available cut-off values for each subscale differentiate between non-cases (subscale score <8), potential cases (subscale score 8-10), and probable cases (subscale score ≥ 11) of anxiety and depression symptoms during the past week. 94 Chronic stress was evaluated using the 12-item screening scale of the Trier Inventory of Chronic Stress. 78 The scale evaluates individual experiences with chronic stressors in everyday life and provides a reliable global measure of perceived stress during the past 3 months. 64 Likert-scale response options are “never” (0), “rarely” (1), “sometimes” (2), “often” (3), and “very often” (4), with a total score ranging from 0 to 48, and higher scores indicating greater perceived presence and frequency of chronic stressors. Patients additionally completed disease-specific measures. The Clinical Activity Index (CAI) 68 consists of 6 items capturing a range of typical IBD symptoms (ie, increase in stool frequency, bloody stools, abdominal pain, temperature due to colitis, extraintestinal manifestations, and the investigator's global assessment of symptomatic state) and 1 item concerning laboratory results (ie, erythrocyte sedimentation rate and hemoglobin). Based on the total sum score, disease activity can be classified into inactive (ie, remission; ≤4), mild activity (5-10), moderate activity (11-17), and high activity (≥18) with a maximum score of 26. 77 The Partial Mayo Score (PMS) 54 is based on 3 non-invasive items (stool frequency, rectal bleeding, global assessment; score range: 0-9; remission ≤1). 79 Finally, to classify chronic pain severity according to von Korff (Grade I, low disability-low intensity; Grade II, low disability-high intensity; Grade III, high disability-moderately limiting; and Grade IV, high disability-severely limiting), 85 an excerpt from the German Pain Questionnaire (Deutscher Schmerzfragebogen, DSF) 65 assessing pain intensity in the past 4 weeks and pain-related disability in the past 3 months was used. The study consisted of 3 sessions. During an initial session, participants filled out the comprehensive questionnaire battery and underwent structural and resting state MRI (day 0; MR-images were used in a larger voxel-based morphometry analysis). 61 Within 4 weeks, participants came back on 2 consecutive days to complete the differential fear conditioning paradigm reported on herein (details below and in Fig. 1 ). The paradigm is empirically validated: Based on first implementations focusing on visceral pain as US 37 and embedded in a growing field of pain-related fear conditioning literature, 30 , 60 it has been refined over the years by incorporating multiple threats from different sensory modalities in healthy individuals, 39 – 41 in experimental models of acute systemic inflammation, 13 , 63 and in patients with chronic pain. 35 , 53 Similar paradigms have also been applied by other research groups, for instance in studies in chronic back pain patients 75 and healthy participants. 76 On study day 1, participants underwent fear acquisition training, directly followed by fear extinction training. On day 2, ie, after overnight consolidation, extinction recall and US re-exposure were conducted. To minimize circadian effects, sessions were conducted between 1 and 7 pm and scheduled at the same time within participants. On day 1 and day 2, saliva samples were acquired upon awakening and before (pre) as well as after (post) the experimental sessions. In addition, blood samples were collected at the beginning of these sessions, and patients were asked to submit a stool sample within 72 hours after day 1. Schematic overview of the differential pain-related fear conditioning paradigm. The timeline depicts the sequence of stimulus ratings as well as saliva and blood sampling. ACQ, acquisition training; AUD, auditory; CS, conditioned stimulus; EXT, extinction training; REC, extinction recall; RE-EXP, re-exposure; US, unconditioned stimulus; VAS, visual analog scale. For differential pain-related fear conditioning, a visual cue (CS + PAIN ) was repeatedly paired with painful heat as US (US PAIN ), whereas another visual cue (CS − ) remained unpaired. To test specificity of findings to pain vs aversiveness, a third visual cue (CS + AUD ) was repeatedly paired with an aversive auditory stimulus as US (US AUD ). For US PAIN , a cutaneous thermal stimulus was applied in the left lower quadrant of the abdomen using a thermode (PATHWAY model CHEPS; Medoc Ltd., Advanced Medical Systems, Ramat Yishai, Israel). Note that for safety reasons, a temperature limit was set to 50°C. For US AUD , an aversive tone with a sawtooth waveform profile and a frequency of 1 kHz was presented binaurally through commercially available headphones (A077B, SoundLAB). Importantly, US PAIN and US AUD were individually calibrated based on perceptual thresholds and matched to unpleasantness for implementation of individual US intensities (ie, temperature or loudness) within a predefined perceptual unpleasantness range of 60 to 80 mm on a 0 to 100 mm visual analog scale (VAS), analog to our previous work. 63 During acquisition training, each CS was presented 12 times, with 9 CS + -US pairings per modality (ie, 75% reinforcement schedule, chosen to induce uncertainty and to ensure robust conditioned responses 37 , 48 , 49 ; Fig. 1 ). All CS + were presented 6 to 12 seconds before US, with CS and US co-terminating (ie, differential delay conditioning). Allocation of a specific CS symbol to a specific US or designation as unpaired CS was counterbalanced across participants. The programming of pairings aimed for an essentially pseudorandomized order but avoided more than 2 successive pairings of 1 type. During extinction training, all CS were presented without any US, using the same pseudorandomized CS sequence as during acquisition. Extinction recall was accomplished after overnight-consolidation and comprised CS presentations without any US, again using the same pseudorandomized CS sequence as during acquisition. Then, participants were unexpectedly re-exposed to 3 US PAIN and 3 US AUD presentations using identical stimulus intensities as during acquisition, before CS were again presented unpaired. Finally, participants were once again re-exposed to US PAIN and US AUD ; this time following the notification that US PAIN and US AUD , respectively, will be applied now to create similar conditions as for baseline US ratings. In all phases, interstimulus intervals consisted of a black screen with a fixation cross with a duration of 8 seconds, and all US durations were 14 seconds. To ensure that learning was successful, using the arrow keys of a standard keypad, participants provided fear ratings on digitized 0 to 100 mm VAS with endpoints labeled “not at all” and “very much” at baseline, after acquisition, after extinction, before extinction recall, and after early as well as late re-exposure. Additionally, as US responses were the main target of our analyses, digitized VAS were used to measure perceived pain intensity, with the anchors “not painful at all” (0 mm) and “extremely painful” (100 mm), and US valence, with the anchors “very pleasant” (−100 mm), “neutral” (0 mm), and “very unpleasant” (+100 mm). These ratings were obtained at baseline, after acquisition training, and after early as well as late re-exposure. All visual stimuli and rating scales were presented with Presentation software (Neurobehavioral Systems, Albany, CA). Blood samples were collected in ethylenediaminetetraacetic acid-treated tubes (S-Monovette, Sarstedt, Nümbrecht, Germany). Plasma was separated by centrifugation (2000 g , 10 minutes, 4°C) and stored at −80°C until analysis. Plasma levels of the candidate pro-inflammatory cytokines tumor necrosis factor (TNF-α) and interleukin-6 (IL-6) were measured by enzyme-linked immunosorbent assays (Human Quantikine ELISA, R&D Systems, Minneapolis, MN). Sensitivity of the assays was 0.11 pg/mL for TNF-α and 0.70 pg/mL for IL-6. Serum concentration of C-reactive protein was measured by ELISA (C-reactive protein ELISA, IBL International, Hamburg, Germany). The sensitivity of the assay was 0.022 mg/dL. White blood cell (WBC) counts were assessed with a hematology analyzer (Sysmex XP-300, Sysmex Europe, Norderstedt, Germany). Erythrocyte sedimentation rate was measured with an automatic blood sedimentation system (Sediplus S 200, Sarstedt). Saliva samples were collected using a synthetic swab (Salivette Cortisol, Sarstedt), which was kept in the mouth for 2 minutes without chewing. Samples were centrifuged at ×1000 g for 2 minutes and stored at −80°C until analysis. Salivary cortisol was measured by ELISA (Cortisol Saliva ELISA, IBL International). Salivary alpha-amylase (sAA) activity was measured by an enzymatic assay (alpha-Amylase Saliva Assay, IBL International). Stool samples were collected with the Enterosan collection kit and directly sent to an external laboratory (Labor LS, Bad Bocklet-Großenbach, Germany) by patients, where levels of calprotectin and lactoferrin were determined. All statistical analyses were performed using R (Version 4.1.1; R Core Team, Vienna, Austria) in RStudio (Version 2021.09.0+351; RStudio, Boston, MA). First, data were visually inspected based on boxplots and tested for normality (Shapiro–Wilk test), homoscedasticity (Levene test), and homogeneity of the variance–covariance matrices (Box M test). As outliers were identified and violations of normality were detected for most variables and timepoints, robust statistical analyses using the WRS2 package 56 were performed instead of classical inferential methods. Yuen t -tests for independent samples were calculated to compare IBD and HC regarding sample characteristics, questionnaire data, endocrine as well as inflammatory markers, and US calibration measures. Although we were mainly interested in US responses during re-exposure to pain following fear learning (ie, on day 2), we first tested successful differential fear conditioning in both groups by analyzing both conditioned and unconditioned responses during fear learning (ie, on day 1). For this purpose, differential fear of conditioned stimuli was computed as individual delta (Δ) score for the CS + PAIN relative to the CS − for each learning phase (ΔCS PAIN fear). Robust mixed analyses of variance (ANOVAs) with time (baseline/after acquisition/after extinction) as within-subject factor and group (IBD/HC) as between-subject factor were performed to test for changes in ΔCS PAIN fear. Similarly, perceived pain intensity and unpleasantness of US PAIN were analyzed by robust mixed ANOVAs with time (baseline/after acquisition) as within-subject factor and group (IBD/HC) as between-subject factor. Post-hoc comparisons were performed using Yuen tests for paired samples, and Bonferroni-corrected P- values are reported where applicable. Subsequently, US responses during re-exposure to pain (ie, on day 2) were analyzed. For this purpose, robust mixed ANOVAs with time (early re-exposure/late re-exposure) as within-subject factor and group (IBD/HC) as between-subject factor were computed for perceived pain intensity and unpleasantness of US PAIN . In addition, to test whether US responses during re-exposure (day 2) relate to pain-related fear learning (day 1), the magnitude of pain-related fear learning was computed as difference score of fear of CS + PAIN after acquisition training and at baseline for each participant, and robust correlations with perceived pain intensity and unpleasantness of US PAIN during early and late re-exposure were calculated in IBD and HC. Winsorized correlation coefficients (ρw) are reported, using the default 20% Winsorization (tr = 0.2). Finally, robust mediation analyses were performed for each group to test whether US PAIN unpleasantness as an indicator of the motivational-affective pain response may mediate the relation between pain-related fear learning and perceived pain intensity. These analyses were performed using the R package robmed, 4 which provides robust mediation procedures based on the fast-and-robust bootstrap methodology for robust regression estimators, yielding reliable results even when normality assumptions are violated. Perceived pain intensity of US PAIN and the magnitude of pain-related fear learning were entered as dependent and independent variable, respectively, and US PAIN unpleasantness was the hypothesized mediator. As aversive auditory stimulation was additionally included in this study, we were able to test specificity of findings to pain vs aversiveness. For this purpose, all analyses were repeated for the auditory modality. To maintain conciseness and our focus on pain, these results are provided in the supplemental material, http://links.lww.com/PAIN/C415 . The alpha level for accepting statistical significance was set at P < 0.05. Results are reported as trimmed mean ± trimmed standard error of the mean unless indicated otherwise. Plots were prepared with the ggplot2 package. 86

From the total of N = 53 participants, we excluded 9 participants due to study discontinuation because of COVID-19 restrictions (N = 3), technical difficulties (N = 3), unsuccessful age-matching (N = 2), or unsuccessful matching of US unpleasantness (N = 1). In addition, data acquired from 1 IBD patient were excluded from analysis due to a change of medication during study participation (ie, initiation of treatment with systemic glucocorticoids). Therefore, the final sample consisted of N = 21 IBD patients with an average disease duration of 13.54 ± 2.46 years (16 female, age: 45.54 ± 3.87 years, BMI: 22.79 ± 0.92 kg/m 2 ; N = 14 currently treated with aminosalicylates) and N = 22 age-matched HC (17 female, age: 46.29 ± 3.63 years, BMI: 22.62 ± 0.65 kg/m 2 ). As expected, the 2 groups did not significantly differ in age or BMI (all P > 0.1). In Table 1 , detailed self-report data on gastrointestinal, pain, and psychological symptoms in both groups are provided. Confirming that our experiments were not conducted in a phase of acute IBD-related pain, there were no group differences in pain severity or functional interference within the last 24 hours before participation. However, in the previous 3 months, IBD patients experienced significantly more gastrointestinal symptoms, including upper and lower abdominal pain, than healthy volunteers. The burden of chronic pain, as indicated by von Korff grading, was comparatively low for the majority of patients (N = 16 Grade I, N = 2 Grade II, N = 3 Grade III, N = 1 Grade IV). Still, pain-related emotions and cognitions were altered in the patient group: Levels of pain-related anxiety as assessed by the PASS-20 were significantly increased, and self-statements related to pain catastrophizing were more prevalent, whereas self-statements related to active coping with pain were less prevalent in patients compared with controls. In line with findings for pain-related anxiety (PASS-20), general subclinical anxiety symptoms (HADS) were increased in IBD patients. By contrast, levels of subclinical depressive symptoms and perceived chronic stress were comparable to those in HC. Sample characterization: gastrointestinal, pain, and psychological symptoms. Questionnaire data from patients with inflammatory bowel disease (IBD) are compared with those of healthy controls (HC) using Yuen t -tests for independent samples. Data are shown as trimmed mean ± trimmed standard error of the mean. * P < 0.05, *** P < 0.001. BPI, brief pain inventory; FPQ, fear of pain questionnaire; HADS, hospital anxiety and depression scale; PASS-20, pain anxiety symptoms scale; PRSS, pain-related self statements scale; TICS, Trier inventory of chronic stress. Although the majority of IBD patients was in clinical remission according to CAI (1.77 ± 0.62, with CAI ≤ 4 for N = 18) and PMS (1.00 ± 0.47, with PMS ≤ 1 for N = 14), one-third of patients was classified to have mild (N = 6) or moderate (N = 1) disease status based on their PMS. In line with this, elevated fecal calprotectin levels (ie, >150 μg/g) and lactoferrin levels (ie, >7.25 µg/g) were observed in some patients (calprotectin: 122.26 ± 33.87, suprathreshold for N = 7 patients; lactoferrin: 6.43 ± 3.53; suprathreshold for N = 7 patients). Although all inflammatory markers derived from blood samples were found to be within reference ranges in both groups (Table 2 ), WBC counts and plasma levels of TNF α and IL-6 were significantly higher in IBD vs HC, indicating systemic low-grade inflammation. Neuroendocrine markers derived from salivary samples (ie, cortisol and sAA activity) did not deviate from normal ranges and were comparable across groups, except for cortisol upon awakening on day 1, which was significantly lower in IBD than in HC (Table 2 ). Sample characterization: immune and neuroendocrine markers. Immune and neuroendocrine markers were derived from blood and saliva samples, respectively. Data from patients with inflammatory bowel disease (IBD) are compared with those of healthy controls (HC) using Yuen t -tests for independent samples. Data are shown as trimmed mean ± trimmed standard error of the mean. * P < 0.05, ** P < 0.01. N = 18. N = 20. N = 21. N = 19. CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; IL, interleukin; sAA, α-amylase activity; TNF, tumor necrosis factor; WBC, white blood cell. Pain thresholds did not significantly differ in IBD patients (43.42 ± 0.94°C) and HC (43.34 ± 1.48°C; P > 0.1), and individual calibration and matching to US unpleasantness were successful, as indicated by comparable baseline unpleasantness ratings for US PAIN (IBD: 51.54 ± 4.21 mm; HC: 52.14 ± 4.96 mm) and US AUD (IBD: 48.00 ± 5.46 mm; HC: 47.71 ± 4.55 mm; all P > 0.1). In line with this, at baseline, there were no significant group differences in perceived pain intensity (US PAIN : 64.85 ± 3.07 mm for IBD, 63.00 ± 3.41 mm for HC; US AUD : 47.00 ± 5.63 mm for IBD, 43.36 ± 9.22 mm for HC; all P > 0.1). Of note, the resulting stimulation intensities applied for US PAIN (IBD: 46.35 ± 0.41°C; HC: 46.33 ± 0.71°C) and US AUD (IBD: 95.08 ± 1.89 dB; HC: 94.79 ± 2.58 dB; all P > 0.1) were also comparable across groups. Analysis of conditioned fear responses (ie, ΔCS PAIN fear ratings) revealed a significant main effect of time ( F [2, 16.64] = 6.43, P = 0.009) but no significant group -related effects ( P > 0.1 for main effect of group and time × group interaction), supporting successful and comparable pain-related fear learning across groups (Fig. 2 A). As expected, post-hoc tests showed that ΔCS PAIN fear ratings, across groups, were significantly higher after acquisition training compared with baseline ( t [26] = 3.08, P = 0.005, ξ ^ = 0.92) and significantly lower after extinction than after acquisition training ( t [26] = 2.83, P = 0.009, ξ ^ = 0.44). Differential fear responses to conditioned stimuli (ie, ΔCS PAIN fear ratings on VAS in mm) are depicted for patients with inflammatory bowel disease (IBD) and healthy controls (HC). Statistical results of robust mixed ANOVAs are indicated. ACQ, acquisition; BASE, baseline; EXT, extinction training; REC, extinction recall; RE-EXP, re-exposure. ** P < 0.01. Similarly, with respect to unconditioned responses during pain-related fear learning, no significant group -related effects were found for perceived pain intensity (Fig. 3 A) or unpleasantness (Fig. 3 B) of US PAIN (all P > 0.05). A significant main effect of time was observed for US PAIN unpleasantness ( F [1, 22.84] = 13.27, P = 0.001 ξ ^ = 0.61) but not for perceived pain intensity of US PAIN ( P > 0.1). Specifically, across groups, US PAIN unpleasantness ratings were significantly lower after acquisition training compared with baseline (Fig. 3 B). Responses to unconditioned stimuli (ie, perceived pain intensity and unpleasantness ratings for US PAIN on VAS in mm) are shown for patients with inflammatory bowel disease (IBD) and healthy controls (HC). Statistical results of robust mixed ANOVAs are indicated. ACQ, acquisition; BASE, baseline; RE-EXP, re-exposure. * P < 0.05, ** P < 0.01. On day 2, when participants were re-exposed to pain, patients differed from healthy controls in their responses to US PAIN , as a significant main effect of group was found in the analyses of perceived pain intensity ( F [1,24.90] = 6.52, P = 0.017, ξ ^ = 0.38) and unpleasantness ( F [1, 23.33] = 6.98, P = 0.015, ξ ^ = 0.31) of US PAIN during re-exposure, respectively. In both analyses, no significant time -related effects were observed ( P > 0.1 for main effects of time and time × group interactions). Hence, irrespective of the timepoint (ie, early and late re-exposure), US PAIN was more painful (Fig. 3 C) and unpleasant (Fig. 3 D) during re-exposure to pain following pain-related fear learning for IBD patients than for HC. In the analysis of conditioned fear responses on day 2, no significant effects were found (ie, ΔCS PAIN fear ratings; all P > 0.1; Fig. 2 B). Correlational analyses demonstrated a significant relation between US PAIN responses during re-exposure and pain-related fear learning that was specific for IBD patients: The magnitude of pain-related fear learning on day 1 (ie, the change in fear of CS + PAIN from baseline to after acquisition training) positively correlated with the perceived pain intensity ( ρw = 0.54, P = 0.017 for early re-exposure; ρw = 0.49, P = 0.034 for late re-exposure) and unpleasantness ( ρw = 0.56, P = 0.013 for early re-exposure; ρw = 0.47, P = 0.040 for late re-exposure) of US PAIN on day 2. Hence, the larger patients' magnitude of pain-related fear learning was on day 1, the more painful and unpleasant the US PAIN was experienced during re-exposure on day 2. By contrast, no significant correlations between the magnitude of pain-related fear learning on day 1 and US PAIN responses during re-exposure on day 2 were found in HC (all P > 0.1). Notably, perceived pain intensity and unpleasantness assessed at baseline did not correlate with the magnitude of pain-related fear learning (all P > 0.1), suggesting learning-related changes in the relation between fear learning and pain perception in IBD patients. Analyses testing the mediating role of US PAIN unpleasantness as an indicator for the motivational-affective pain response revealed that the relations between the magnitude of pain-related fear learning (ie, change in fear of CS + PAIN from baseline to after acquisition training on day 1) and perceived pain intensity of US PAIN during early and late re-exposure (on day 2) observed in IBD patients were mediated by the unpleasantness of US PAIN during early and late re-exposure (on day 2), respectively. Bootstrapped coefficients and significance levels are provided in Figure 4 . Direct effects of the magnitude of pain-related fear learning on perceived pain intensity of US PAIN during early and late re-exposure were not significantly different from 0 in IBD, meaning that the magnitude of pain-related fear learning was related to patients' perceived pain intensity of US PAIN during re-exposure only via the indirect path through US PAIN unpleasantness. Hence, in the typology of mediations of Zhao et al., 93 indirect-only mediation was found in IBD. By contrast, in HC, neither direct nor indirect effects of the magnitude of pain-related fear learning on perceived pain intensity of US PAIN during early re-exposure existed (ie, no-effect non-mediation). For late re-exposure, only a direct effect of the magnitude of pain-related fear learning on perceived pain intensity of US PAIN was found but no mediation by US PAIN unpleasantness (ie, direct-only non-mediation). Bootstrapped coefficients and significance levels based on mediation analyses performed for US PAIN responses during (A) early re-exposure and (B) late re-exposure for patients with inflammatory bowel disease (IBD) and healthy controls (HC). CS, conditioned stimulus; US, unconditioned stimulus. * P < 0.05, ** P < 0.01, *** P < 0.001.

Recurring pain is a clinical challenge in IBD. 89 Despite progress in understanding IBD pathophysiology and brain–gut interactions, 15 the mechanisms underlying pain persisting during clinical remission remain elusive. Psychological factors, especially altered emotional reactivity to pain, have been proposed as key contributors, 14 , 80 but experimental evidence is scarce. Focusing on pain-related fear as a fundamental emotional process implicated in pain chronification 71 and building on our prior work demonstrating altered neural processing during pain-related fear learning in IBD, 53 this study elucidated the relevance of fear learning and memory processes for subsequent pain perception in patients with quiescent IBD. Our principal finding is that patients with IBD experienced significantly higher pain intensity and unpleasantness upon re-exposure to pain compared with healthy volunteers. As hypothesized, this enhanced pain perception correlated with the magnitude of pain-related fear learning the day before, making this the first study to demonstrate fear-induced hyperalgesia in quiescent IBD. Mediation analyses further revealed that pain unpleasantness fully mediated the relationship between fear learning and perceived pain intensity. Hence, as expected, the emotional valence assigned to the pain experience played a critical role. These results provide novel evidence that emotional learning and memory processes shape pain perception in IBD, supporting conceptual models in which heightened emotional reactivity to pain contributes to symptom persistence. 14 , 80 Perceptual changes following pain-related fear learning reflect the active, inferential nature of pain perception, in which prior information and expectations about future perception guide interpretation of sensory input. 82 , 87 Beyond explicit negative expectations, pain modulation through central or descending modulatory systems involves implicit associative learning processes, 8 , 9 as illustrated by nocebo phenomena. 17 Our findings newly demonstrate the relevance of emotional learning and memory, identifying fear-induced hyperalgesia as a potential mechanism in persistent IBD-related pain. The observed effects likely reflect alterations in central pain modulatory circuits that amplify nociceptive signals in response to learned fear, consistent with the emerging framework of nociplastic pain 44 in chronic inflammatory conditions. 26 The International Association for the Study of Pain introduced the term nociplastic pain in 2017, defining it as “pain that arises from altered nociception despite no clear evidence of actual or threatened tissue damage causing the activation of peripheral nociceptors or evidence for disease or lesion of the somatosensory system causing the pain.” This third mechanistic descriptor complements nociceptive and neuropathic dimensions. 44 Clinical criteria are still in development, 42 , 43 but the term highlights the role of central adaptations in chronic pain, calling for conceptual refinement and research to clarify if and how inflammation-driven nociceptive pain mechanisms may interact with (or exist in parallel with) nociplastic pain mechanisms. Although this study did not include neuroimaging measures, the behavioral patterns align with prior brain imaging data acquired in patients 53 , 61 , 62 , 72 and animal models 20 , 33 , 57 , 73 showing widespread (pain-related) central alterations in IBD. Together, these findings point toward nociplastic mechanisms as key contributors to persistent pain in quiescent IBD, in line with work in other inflammation-related chronic pain conditions such as rheumatoid arthritis 74 and endometriosis. 19 Behavioral measures of fear acquisition and extinction were largely unaltered in patients compared with controls. As hypothesized, both groups learned and extinguished conditioned fear to pain-predictive cues comparably, replicating our earlier findings of intact fear learning in IBD despite altered neural activity during these processes. 53 This aligns with results from experimental endotoxemia—a model of acute systemic inflammation in healthy individuals 12 —where fear learning was similarly intact, but pain re-exposure elicited enhanced responses. 65 Together, these findings suggest that inflammation may leave a “fingerprint” within the fear network, influencing post-inflammatory pain perception, and point toward pain-related central adaptations rather than exaggerated fear acquisition as the underlying mechanism. Notably, this differs from other chronic pain disorders such as IBS or chronic low back pain, where impairments in learning and overgeneralization of fear have been reported. 31 , 35 , 60 , 75 Furthermore, in IBS patients, no indications for altered pain responses after fear learning were found. 35 These distinctions may reflect disease-specific pathophysiological mechanisms, corroborating our earlier conclusions based on direct comparisons of IBD and IBS cohorts with respect to neural mechanisms underlying pain and fear 53 , 61 , 62 and underscoring the close interaction between immune function and affective processes. 29 The emotional valence, referring to the subjective value assigned to sensory stimuli (pleasant vs unpleasant), 66 , 70 plays a key role in the effects observed in our study, as the relationship between pain-related fear learning and perceived pain intensity was fully mediated by pain unpleasantness. The amygdala and interconnected corticolimbic regions, which are crucial for coding valence, 18 , 66 are reportedly sensitive to peripheral inflammation 22 , 45 and show functional alterations in both IBD patients 2 , 3 , 72 and animal models. 20 , 33 , 57 Although affective-motivational and sensory-discriminative responses are dissociable, the emotional valence of pain is usually closely linked to perceived pain intensity. 67 Consistent with this, our mediation analyses showed that perceived pain intensity was associated with pain unpleasantness in both IBD patients and healthy individuals. However, the relationship between perceived pain intensity during re-exposure and the magnitude of pain-related fear learning was only found in patients and was fully mediated by pain unpleasantness during re-exposure. To our knowledge, no previous studies have examined the link between pain-related fear learning and the valence of pain in patients with chronic pain or healthy individuals. Although it is well-established that pain-predictive cues acquire negative valence, i.e., become more aversive or unpleasant, 60 and may even be perceived as more intense over the course of pain-related fear conditioning, 91 our results highlight the need for further investigation into how pain-related fear learning shapes the emotional valence of pain itself, especially in vulnerable individuals. This study's strengths include the translational 2-day fear conditioning paradigm allowing rigorous assessment of learning, memory, and re-exposure processes; and the inclusion of well-characterized IBD patients with detailed clinical, psychological, immunological, and endocrine profiling. This minimized confounding by inflammatory bouts and putative effects of medical treatments required during phases of disease exacerbation. Additionally, the combined assessment of nociceptive and non-nociceptive aversive stimuli can provide important insights into specificity: Unexpectedly, we observed that enhanced responses upon re-exposure were not restricted to pain but extended to aversive tones. This finding expands previous reports of heightened emotional responses to diverse aversive stimuli and stressors in IBD, 1 – 3 , 72 underpins the notion of enhanced central fear network reactivity, 53 and emphasizes the need to consider altered emotional reactivity more broadly in this population in mechanistic exploration. This is further supported by evidence of hypersensitivity to a range of non-nociceptive environmental (ie, sensory) stimuli, including lights and sounds, in other nociplastic pain conditions, 24 , 36 suggesting that it may represent a transdiagnostic feature. Limitations of our study include a small sample size and the relatively low chronic pain burden and psychological distress in our cohort. These factors limit generalizability and highlight the need for studies in more diverse IBD populations to clarify the impact of gastrointestinal symptom profiles and psychological comorbidities. Additionally, it remains uncertain whether our findings from quiescent IBD extend to patients with active disease. Because some analyses are correlational, we cannot infer causality; however, the temporal design and the absence of a significant correlation between baseline pain perception and pain-related fear learning support mechanistic interpretations. Research in active disease is particularly warranted, given evidence that psychological processes, including mood states, influence pain-related responses in experimental models of inflammation. 12 , 63 Acute inflammation is linked to corticolimbic brain regions involved in affective processing, 45 suggesting that similar mechanisms may contribute during active disease phases. Moreover, understanding how repeated episodes of acute inflammation drive long-lasting central adaptations represents an important avenue for future research. Persistent pain is a disabling symptom in IBD that remains refractory to anti-inflammatory treatment in a subset of patients. 89 Our results suggest that pain-related fear learning may contribute to pain persistence beyond inflammatory bouts. The fear-avoidance model of chronic pain, 84 recently integrated into a novel psychobiological model of disorders involving the gut–brain axis, 30 provides a strong theoretical basis for targeting conditioned fear in conditions involving recurring abdominal pain, including IBD. Although exposure-based CBT is effective in IBS, 6 , 38 evidence in IBD is mixed, 11 often improving psychiatric comorbidities and quality of life but not bodily symptoms. 71 No evidence-based exposure therapies currently exist for IBD. Notably, even in our relatively psychologically healthy patient cohort, subtle dysregulation of stress systems and inflammatory markers was observed. This has previously been reported in IBD 23 and indicates vulnerability that might be exacerbated in patients with psychiatric comorbidities or high chronic stress. These subgroups warrant focused investigation, as they may be at higher risk for fear-induced hyperalgesia and poor outcomes. Personalized, mechanistically informed treatments integrating psychological and physiological targets could fill the current therapeutic gap in IBD-related pain. 34

Abdominal pain is a prevalent symptom of inflammatory bowel disease (IBD), 10 significantly impacting patients' psychological health, quality of life, and healthcare utilization, even during disease remission. 16 , 27 , 28 , 80 As pain often persists beyond acute gut inflammation and remains a clinical and mechanistic challenge, 89 IBD is increasingly recognized as a pain disorder. 7 , 10 Besides sensitization of visceral afferent neurons by inflammation, emotional reactivity to the affective dimension of pain may be altered. 14 , 80 Yet, how emotional responses modulate IBD patients' perception of experimentally induced acute pain remains poorly understood. A critical pain-related emotion is fear, enabling protective behaviors like withdrawal or avoidance. 55 Through classical conditioning, preceding or coinciding stimuli become associated with pain, causing anticipatory fear. Maladaptive fear learning contributes critically to pain chronification 84 and has been extensively studied in chronic pain conditions, 60 including irritable bowel syndrome (IBS) as a disorder of gut–brain interaction. 35 , 51 In the first study on pain-related fear learning in IBD patients, we recently found normal anticipatory fear responses but altered neural activations within the central fear network during conditioning. 53 To test whether these central alterations may reflect changes in pain processing, this study aimed to elucidate whether fear learning shapes subsequent pain perception in IBD. Direct modulation of pain perception by anticipatory processes is a well-established phenomenon, eg, in nocebo effects where negative expectations worsen bodily symptoms like pain, 17 and is targeted in cognitive behavioral therapy (CBT) protocols. 88 However, most studies suggesting fear learning may enhance pain perception, including intensity and unpleasantness, 21 , 91 are limited to healthy volunteers. 81 , 90 , 92 In healthy individuals with an experimentally-induced acute episode of systemic inflammation, 63 we recently showed enhanced pain unpleasantness after fear learning. Understanding these psychological pain mechanisms in conditions with relapsing-remitting inflammation and persistent pain like IBD may inform personalized treatments including psychological interventions based on the fear-avoidance model 84 and exposure therapy. 38 Therefore, we implemented an established 2-day differential fear conditioning paradigm 13 , 35 , 39 – 41 , 53 , 63 in patients with quiescent IBD and healthy controls (HC). On study day 1, fear acquisition and extinction in response to conditioned stimuli (CS) paired with nociceptive (thermal pain applied to the abdomen) and non-nociceptive (aversive tones) unconditioned stimuli (US) were assessed. Responses to US re-exposure were tested on study day 2. Our primary focus was on pain intensity and unpleasantness ratings upon re-exposure to pain, chosen as novel, clinically relevant outcomes allowing us to explore the putative impact of conditioned fear on subsequent pain perception. We expected no group differences in CS-related fear ratings during conditioning, consistent with prior work, 53 but hypothesized increased pain intensity and unpleasantness in IBD upon re-exposure, correlating with the magnitude of pain-related fear learning. Building on theories of pain-related altered emotional reactivity in IBD, 14 , 80 we explored pain unpleasantness as a mediator. While assuming specificity to pain, 53 , 63 exploratory analyses considered responses to the aversive auditory US to evaluate potential generalized emotional reactivity.

Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/C415 .

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.