Associations between sickness behavior, but not inflammatory cytokines, and psychiatric comorbidity in chronic pain

Objectives: Approximately one in five adults experiences chronic pain, often in co-occurrence with depression, insomnia, anxiety, and lower self-rated health. Elevated levels of cytokines, e


Introduction
Chronic pain, as outlined in the ICD-11 classification system refers to persistent pain lasting beyond three months (Nicholas et al., 2019).The International Association for the Study of Pain (IASP) defines pain as an unpleasant sensory and emotional experience that may or may not be linked to actual or potential tissue damage (Raja et al., 2020).Around one-fifth of the adult population experiences chronic pain, often together with depression, anxiety, insomnia (Breivik et al., 2006), obesity (Okifuji and Hare, 2015), lower self-rated health (Mäntyselkä et al., 2003) and experiences of sickness (Jonsjö et al., 2020).Socio-demographic factors associated with chronic pain include older age, female gender, and lower socioeconomic status (SBU, 2006).The high prevalence, the complex interplay of different symptoms, and the inefficiency of medical interventions (Turk, 2002) make chronic pain a condition with a possible negative impact on daily functioning and quality of life for a large proportion of the population (Zetterqvist et al., 2017;Breivik et al., 2006).
The etiology of chronic pain is in many cases unclear, but inflammatory mechanisms have been suggested as possible contributing factors in the development and maintenance of chronic pain (Ren and Dubner, 2010;Grace et al., 2014).Higher levels of pro-inflammatory markers, e.g.IL-6 and IL-8 have been associated with greater pain intensity in several pain types (Koch et al., 2007;DeVon et al., 2014).In chronic pain, the literature is not unanimous whether certain cytokines are elevated.The cytokines tumor necrosis factor alpha (TNF-α), interleukin 6 (IL-6), interleukin 8 (IL-8), and interleukin 10 (IL-10), have been identified in meta-analyses, systemic reviews, and a literature review including patients with e.g.fibromyalgia (Üçeyler et al., 2011;Andrés-Rodríguez et al., 2020), non-specific low back pain (Morris et al., 2020;van den Berg et al., 2018) and neuropathic pain (Hung et al., 2017).However, one meta-analysis illustrated that the majority of the cytokines in the included studies were not different between patients and controls (Üçeyler et al., 2011).The discrepancies between studies highlighting elevated cytokine levels in patients with chronic pain and those showing no significant differences underscore the complexity of chronic pain conditions and the challenges in synthesizing heterogeneous research findings.The discrepancies in cytokine findings between studies could for example be attributed to sample variability, methodological variations, publication bias or heterogeneity of chronic pain.
Increased levels of inflammatory cytokines have also been found in several previously mentioned conditions related to chronic pain, such as major depressive disorder (Haapakoski et al., 2015;Liu et al., 2012).Higher CRP levels have been associated with the severity of depressive symptoms, suggesting a possible subgroup of depressed patients that experience inflammation-related depression (Köhler-Forsberg et al., 2017).The literature is inconclusive when examining the relationship between inflammatory biomarkers and sleep deficiency.Furthermore, some research has indicated that sleep deficiency may be contributing to increased inflammatory activity (Irwin et al., 2016), although another study noticed a lack of relationship between sleep duration and CRP (Taheri et al., 2007).Self-rated health has been associated with interleukin 1 beta (IL-1β), interleukin 1 receptor antagonist (IL-1ra), TNF-α (Lekander et al., 2004) and erythrocyte sedimentation rate (ESR) (Warnoff et al., 2016).Results for the relationship between anxiety disorders and inflammatory biomarkers are also however inconclusive (Renna et al., 2018;Furtado and Katzman, 2015).Research including well-designed longitudinal studies and meta-analyses that address potential sources of bias, is needed to clarify the role of cytokines in chronic pain and conditions related to chronic pain.
Hart characterized sickness behavior as "a coordinated array of adaptive behavioral adjustments that emerge in individuals during the progression of an infection" (Hart, 1988).Sickness behavior is linked to inflammatory activity, comprising several behavioral responses, such as worsened mood, anxiety, anhedonia, malaise, fatigue, reduced social activity, and increased pain sensitivity after inflammatory activity (Dantzer et al., 2008).More persistent sickness behavior may add to interfering symptoms in patients with chronic pain (Dantzer et al., 2008).Interestingly, patients with chronic pain display comparable levels of sickness behavior as healthy individuals with experimentally induced inflammation (Jonsjö et al., 2020).
In summary, the existing literature tentatively suggests a relationship between inflammatory biomarkers and depression, insomnia, self-rated health, sickness behavior as well and pain intensity.However, the strength and implication of these relationships in patients with chronic pain are still unclear.More empirical data on the complex relationships between these factors and inflammatory markers may enhance the understanding of key mechanisms in the development, maintenance, and treatment of chronic pain.

Aims
The aim of the study was to investigate the interrelationships between levels of low-grade inflammatory biomarkers (TNF-α, IL-6, IL-8, IL-10, CRP, and ESR) and levels of depression, anxiety, insomnia, pain intensity, self-rated health, and sickness behavior in adult patients with chronic pain.We hypothesized that there would be a positive correlation between low-grade inflammation, pain intensity, and comorbid symptomatology.

Participants and procedure
The cross-sectional data presented here were collected at baseline in a larger study evaluating treatment effects of contextual behavioral therapy (Karshikoff et al., 2022;Rickardsson et al., 2021).The included participants and the methods employed for data collection have been described earlier (Karshikoff et al., 2023).Participants were consecutively recruited between 2016 and 2018, via referrals from primary and tertiary care units in Stockholm County, and an advertisement in local newspapers.Potential participants were eligible for inclusion if they: Were ≥ 18 years of age; presented with a pain duration > 6 consecutive months (which negatively impacted daily functioning and was not expected to be alleviated by medical intervention); were able to communicate in Swedish; and had been on stable medication for the last two months.Potential participants were excluded if they participated in a simultaneous treatment based on cognitive-behavioral therapy (CBT) or presented with severe psychiatric co-morbidity that required acute assessment or treatment (e.g., suicidal ideation, psychotic symptoms), or if a spontaneous improvement could be expected.Additional exclusion criteria for phlebotomy were having given birth within the last year, breastfeeding, pregnancy, and hemophilia.Psychiatric conditions were assessed using a modified version of the Mini International Neuropsychiatric Interview version 5 (Sheehan et al., 1998).Participants completed the self-report questionnaires in conjunction with blood sampling.The recruitment was concluded earlier than anticipated due to logistical circumstances in the clinical setting.The study was registered at ClinicalTrials.gov,Reg no.NCT03272893.

Demographics and clinical data
Data on age, gender, demographics, medications, pain duration, pain localization, and co-occurring symptoms (sickness feeling, fatigue, concentration difficulties, memory deficits, sensitivity to stress, and recurring fever) were collected at study start.Levels of sickness behavior, pain intensity, self-rated health, insomnia, depression, and anxiety were assessed using self-report questionnaires with adequate validity and reliability, presented in more detail below.

Pain intensity
Pain intensity during the last week was assessed using a numeric rating scale (NRS) ranging from 0 ("no pain") to 10 ("worst imaginable pain").

Depressive symptoms
The Patient Health Questionnaire-9 (PHQ-9) was used to measure depressive symptoms.The frequency of depressive symptoms for the past two weeks is rated on a 4-point Likert scale from 0 ("not at all") to 3 ("nearly every day").The questionnaire includes 9 items, and the total scores range from 0 to 27.The questionnaire has good diagnostic validity and internal consistency (Kroenke et al., 2001).The questionnaire has been translated into Swedish.In this sample, Cronbach's alpha was .84.

Anxiety
The Generalized Anxiety Disorder-7 (GAD-7) questionnaire was used to measure anxiety.The frequency of anxiety symptoms is scored for the last two weeks on a 4-point Likert scale from 0 ("not at all") to 3 ("nearly every day").The questionnaire includes 7 items, and the sum scores range from 0 and 21.The questionnaire has a strong criterion and construct validity, excellent internal consistency, and good test-retest reliability (Spitzer et al., 2006).The questionnaire has been translated into Swedish.In this sample, Cronbach's alpha was .85.

Insomnia
The Insomnia Severity Index (ISI) was used to measure insomnia (Bastien et al., 2001).ISI is a questionnaire with seven items on a Likert scale ranging from 0 ("not at all") to 4 ("very much").Sum scores range from 0 to 28.ISI has demonstrated adequate psychometric properties and has been validated in Swedish for use with chronic pain patients.In this sample, Cronbach's alpha was found to be .90.

Self-rated health (SRH-5)
A self-rated health self-assessment with five response options (SRH-5) was used.In the questionnaire, the participants were asked "In general, would you say your health right now is" and the answers were rated on a five-point Likert scale from "very good" to "very poor".SRH shows good reproducibility, reliability, and strong concurrent validity (DeSalvo et al., 2006).

Sickness behavior
The Sickness Questionnaire (SicknessQ) (Andreasson et al., 2018), was used to measure sickness behavior and sickness symptoms.Ten items are rated on a 4-point scale ranging from (0) "disagree" to (3) "agree" and the total score ranges from 0 to 30.SicknessQ has previously in a sample of patients in Swedish primary care been found to have adequate internal consistency and construct validity (Andreasson et al., 2018).The questionnaire has also been validated in Swedish in patients with chronic pain, with adequate structural validity as well as adequate internal consistency (Åström et al., 2022).In this sample, Cronbach's alpha was .83 for the total scale.

Inflammatory markers
Max 50 ml of blood (non-fasting) per participant was sampled in the morning (8-12 AM) and processed within two hours.CRP levels (assessed with a clinical high-sensitivity CRP test), ESR, white blood cells counts with differentials were analyzed immediately through the hospital's clinical routine laboratories.At CRP levels above 10 mg / L, a physician notified participants.Plasma samples for cytokine analysis were processed via the PreBio process linked to the Stockholm Medical Biobank and the Study Center for Laboratory Medicine, KS.The Olink Proseek Multiplex Inflammation panel was used for analysis, which provides a multiplex immunoassay that allows analysis of 92 inflammation-related protein biomarkers (Olink, Uppsala, Sweden; https://www.olink.com/products/inflammation/),as part of another substudy (Karshikoff et al., 2023).This multiplexing level is attained with a proprietary Proximity Extension Assay (PEA) technology.Data are expressed as normalized protein expression (NPX), which can be used for statistical analysis and express relative quantification between samples.For this study, cytokines were chosen in a hypothesis-driven approach based on prior studies and meta-analyses showing relationships between pro-and anti-inflammatory cytokines and the outcomes studied in this article (e.g.Üçeyler et al., 2011;Andrés-Rodríguez et al., 2020).Plasma levels of TNF-α, IL-6, IL-8 and IL-10 were thus included in the current analyses.All selected markers had adequate detectability; no cytokine levels were lower than the threshold of detection.The cytokines were standardized using Z-score standardization, i.e. a mean = 0 and a standard deviation = 1.

Statistical analyses
Analysis of missing data were conducted by visual inspection and heatmaps.Incomplete items were excluded from the analysis.The possible differences between demographic background variables, selfreport questionnaires, and levels of inflammatory biomarkers of the group recruited via referrals and the group recruited via advertising, were analyzed by using Welch's t-test and χ 2 test.

Descriptive statistics
Descriptive statistics were used to describe the participants on a group level in regards to means, standard deviations, and frequencies of demographic factors (such as age and gender), background variables (such as pain duration and the use of pain medication), self-reported questionnaire assessments and the levels of the included inflammatory biomarkers.

Assumptions and inferential statistical analyzes
Assumptions of normality and linearity were inspected using scatterplots and histograms in combination with the Shapiro-Wilk test.Absolute skewness values >1 and kurtosis values >1 were assessed as criteria for non-normally distributed data.Normality of residuals and independence were inspected using scatterplots and quantile-quantile (QQ) plots.The homogeneity of residual variance (homoscedasticity) was examined with scatterplots of residuals versus predicted values.Scatterplots were used to examine the independence of the residual error terms.Potential outliers were examined in these described plots, using the cut-off of three standard deviations for the mark of excluding outliers.In order to assess multicollinearity among predictor variables, we calculated the Variance Inflation Factor (VIF) for each variable in the regression models.A VIF threshold of 5 was selected to identify potential multicollinearity issues (Marcoulides and Raykov, 2019).
To investigate the relationships between levels of depression, anxiety, insomnia, self-rated health, sickness behavior, pain intensity, demographic variables (age, gender), background variables (pain duration and BMI), and inflammatory biomarkers bivariate Spearman correlations were calculated.We considered correlations between .10-.29 as weak; correlations between .30-.49 as moderate; and correlations between .50-.89 as strong (Cohen, 1977).Bonferroni adjustment to the correlation analyses was also used as a sensitivity test.
Subsequently, we performed two multiple linear regression analyzes, with insomnia and depression as dependent variables (continuous scores) to further evaluate the contribution of the included independent variables, as well as the explained variance in the two models.The independent variables were included based on theoretical relevance and significant findings in the correlation analysis.

Software
All statistical analyzes were performed using R version 4.1.2(R Core Team, 2019).The Psych package (Revelle, 2020) was used for analyzing J.L.M. Åström Reitan et al.
Cronbach's alpha, assumption testing as well and linear regression analysis.

Participants
One hundred thirteen participants consented to study participation, and of these 24 did not meet inclusion criteria, or met exclusion criteria.Of the 89 remaining participants; three were excluded because they had > 14 days between blood sampling and questionnaire completion; three because blood samples were obtained past noon; two because biomarker data were missing (one missing cytokine data and one missing hospital tests); and one because the CRP value indicated ongoing infection (>40 mg/L).In total 80 participants were included in the analysis.The mean number of days between blood sampling and questionnaire completion was 2.6 days (SD = 2.8) for the cytokines and 2.4 days for the hospital tests (SD = 2.8).Regarding demographic variables, background variables, pain intensity, self-report questionnaires, and inflammatory biomarkers, 10 participants (12.5 %) had certain missing data.Visual inspection and heat maps indicated that missing data were missing at random.All analyses were performed using listwise deletion, which excludes incomplete items.
The participants recruited through internet advertisement were significantly older, with a mean age of 55.0 compared to 40.0.Additionally, they had a significantly longer mean pain duration (19.5 versus 7.0) and had significantly higher education.The participants who were recruited via referral rated significantly higher scores on depressive symptoms (p < .05)and anxiety symptoms (p < .05)than the group recruited via advertisement.However, there were no differences in inflammatory biomarkers between the two groups.
Participants' had a mean age of 50.8 (SD = 14.7) years, were mainly women (72.5 %), and reported a mean pain duration of 16.7 (SD = 13.3)years.The mean BMI was 25.4 (SD = 4.5) in the sample, and almost all participants were prescribed relevant medicine.Back pain was the most rated pain localization (70.0 %) and 31.3 % indicated entire body pain, 13.8 % reported one pain localization, and 86.3 % multiple pain localizations in the sample.Only 5 % of the sample rated that they did not experience other symptoms than pain.Being easily tired, stresssensitive, and having concentration and memory problems were reported among the majority of participants.
A majority of the sample (63.8 %) fulfilled the criteria for a psychiatric diagnosis according to the MINI screening.Based on self-report, thirty-two participants (45.0 %) rated depressive symptoms (PHQ-9), which indicated that further investigation was recommended, 15 (21.1 %) participants rated moderately severe anxiety symptoms (GAD-7), and 39 (52.0 %) participants rated insomnia on moderately severe and clinically significant levels (ISI).Further characterization of demographic and background variables, pain variables, and related symptoms as well as psychological co-morbidity is available in Table 1.Table 2 provides a summary of the mean, standard deviation, minimum as well as maximum values for the clinical parameters.
CRP in this sample had a range from .2 to 14 with a mean of 2.2 (SD = 2.7).A reference for these values is from healthy young adults, where the median concentration of CRP was .8mg/l and the 99th percentile was 10 mg/l (Shine et al., 1981).A histogram (Fig. 1) provides a visual representation of the dataset's distribution.

Correlational analyzes
Analyzes showed that there were significant correlations between: Insomnia and CRP (r s =.26, p =.02); sex and ESR (r s = .29,p = .01);age and IL-6 (r s = .29,p = .01)and IL-8 (r s = .30,p = .01);BMI and IL-6 (r s = .50,p < .001),CRP (r s = .63,p < .001)and ESR (r s = .42,p < .001).Using an adjusted threshold of significance (0.003) after Bonferroni correction, only the latter three correlation analyses were significant.Pain duration, pain intensity, self-rated health, anxiety, depression, and sickness behavior did not correlate significantly with any of the biomarkers.Detailed test statistics of all correlations can be found in the correlational matrix in Table 3.
Evaluation of assumptions before correlation analyzes indicated that the data was not normally distributed.Due to non-normality, bivariate Spearman correlations were used.

Regression analyses
The visual observation of residual QQ (quantile-quantile) plots indicated sufficient normality and the scale location plots indicated homoscedasticity.The VIF values were below the predetermined threshold of 5 for all predictor variables in the regression models, indicating low levels of collinearity among the predictors.Further, no outliers were > three standard deviations, and therefore included in the analyses.In the linear regression models with the dependent variables depression and insomnia, the independent variables were age, BMI, CRP, TNF-α, IL-6, pain intensity last week, anxiety, and sickness behavior.In the regression model with depression as the dependent variable, sickness behavior (β =.32) and anxiety (β =.40) were positively significant, explaining 49 % of the total variance in depression.In the regression model with insomnia as the dependent variable, sickness behavior (β =.37) was positively significant, explaining 31 % of the total variance in insomnia.Please see Table 4 for additional test statistics.

Main findings
In this sample, cross-sectional relationship between inflammatory biomarkers and the included symptom variables was absent or weak.Therefore, our hypothesis that there was a positive correlation between low-grade inflammation, pain intensity, and comorbid symptomatology was not supported in this study.Participants rated a relatively high overall symptom burden and sickness behavior.Also, sickness behavior was significantly associated with depression and insomnia, whereas inflammatory biomarkers were not.

Inflammatory status and associations to symptom variables
We observed that the clinical routine proteins overall fell within the normal range.However, some participants did exhibit values higher than the clinical reference interval for CRP, ESR, leukocytes, lymphocytes and neutrophils.The lack of cross-sectional connections between inflammatory markers and symptom variables could be due to several factors.First, networks of inflammatory agents may be a more proper level of analysis as compared to individual markers, or that the included inflammatory markers are not the ones of interest.In a previous study conducted by our group (Karshikoff et al., 2023), ten specific inflammatory proteins (STAMBP, SIRT2, AXIN1, CASP-8, ADA, IL-7, CD40, CXCL1, CXCL5, and CD244) exhibited significant correlations with the severity of anxiety and depressive symptoms reported by the patients with chronic pain.Secondly, our approach in this study was to analyze the commonly studied cytokines and clinical variables separately.Other research goals may benefit from compound scores, such as several cytokines representing inflammatory activity (Karshikoff et al., 2022;Lasselin et al., 2016), or several survey measures representing psychological comorbidity (Karshikoff et al., 2023).Here, CRP levels could represent a more general inflammatory activity, as a compound score would.Furthermore, publication bias might exist favoring articles that highlight a correlation between symptoms and inflammatory markers, potentially leading to an oversight of non-findings.We have previously shown that the self-rated sickness-behavior can be as severe in clinical populations as in experimental settings, despite presumably different levels of blood inflammatory marker levels (Jonsjö et al., 2020).

Sample characteristics and symptom burden
The overall symptom burden in the sample can be illustrated by high mean pain duration and numerically higher ratings of depressive symptoms, anxiety, and insomnia than expected in the general population.The majority of participants fulfilled the criteria for a psychiatric diagnosis.In the current sample, 12.5 % fulfilled the criteria for generalized anxiety, which is substantially higher compared to the estimated prevalence of 2.8-8.5 % in general medical practice, and 1.6-5.0% in, the general population (Spitzer et al., 2006).Similarly, 20.0 % fulfilled criteria for a depression diagnosis, contrasting the 5.2 % point prevalence of major depression in the Swedish adult population (Johansson et al., 2013).Regarding insomnia, 52.0 % of participants rated clinically significant on insomnia (ISI), a considerably higher percentage than the estimated 24.6 % in the general population that report symptoms of insomnia (Mallon et al., 2014).Although the participants recruited after referral rated significantly higher on depression and anxiety compared to the participants recruited via advertisement, the pattern in this sample as a whole regarding depression and anxiety is congruent with other studies presenting participants with chronic pain, indicating representativeness of the current sample (Gerdle et al., 2019).Higher BMI is connected with higher CRP concentrations, proposing a state of low-grade systemic inflammation in overweight and obese persons, which may contribute to pain, a relationship which has also been reported in fibromyalgia patients (Okifuji and Hare, 2015).In our study sample, BMI had a medium-strong association with several inflammatory markers.Eighteen percent of the participants in this study were classified as obese, whereas the corresponding figures for obesity have been estimated to be 14.8 %, and 11.0 % for men and women respectively in the general Swedish population (Neovius et al., 2006).BMI correlated significantly and medium-strong association with IL-6, CRP, and ESR, indicating that BMI has a stronger association with inflammatory biomarkers than the other variables investigated.

Chronic pain and sickness behavior
In the present sample, inflammatory biomarkers did not correlate with the participants' perceived sickness behavior or their self-rated health.Notably, participants in this sample rated sickness behavior on a similarly high level as previously shown, on par with healthy adults during an acute inflammatory response resulting from injection with lipopolysaccharides (LPS), and even higher than primary care patients, some of them seeking consultation due to infections (Jonsjö et al., 2020).Also, almost all participants in this sample experienced other symptoms beyond pain, similar to sickness symptoms, such as being easily tired and experiencing concentration problems.Notably, sickness behavior specifically measured with SicknessQ, explained significant variance in both depression and insomnia, indicating that sickness behavior, i.e. the symptoms of a sickness response, is a relevant level of analysis in insomnia and depression in participants with chronic pain to a greater extent than inflammatory biomarkers.
Sickness behavior is believed to have evolved to facilitate recovery from an acute illness (Dantzer et al., 2008).However, from a learning theory perspective, there are several routes for symptoms to persist.Selective attention may enhance the awareness of nonthreatening bodily sensations that would otherwise go unnoticed, intensifying and prolonging symptoms.This, in turn, could increase the frequency of associative learning occurrences (De Peuter et al., 2007).Deficiencies in associative learning, such as overgeneralization, diminished safety learning, and reduced differential learning, may serve as transdiagnostic vulnerability markers (Boddez et al., 2012).Additionally, the expectancy effect in pain and sickness behavior is noteworthy, as evidenced by pain-intensity ratings indicating that the same stimuli were perceived as less intense when a lower pain intensity was expected (Koyama et al., 2005).Whether stemming from a prior sickness response or not, the patients' perceived symptoms hold clinical relevance, as indicated by a study displaying that elevated levels of sickness behavior correlated with lower self-rated health and diminished health-related functioning (Jonsjö et al., 2020).These associations may signify a vicious cycle wherein perceived sickness symptoms are impeding daily life, followed by a lowered activity level and impaired functioning which sequentially further impair health.The lowered activity level can be due to factors such as illness behavior and sick role, which can influence the manifestation of symptoms in patients.The relationship between chronic pain and psychiatric co-morbidities is most likely bidirectional.It has for example been shown that depression exacerbates pain, but that pain contributes to depression as well, in musculoskeletal pain (Magni et al., 1994).Furthermore, the presence of multiple psychiatric disorders in patients with chronic pain amplifies disability (McWilliams et al., 2003).

Strengths and limitations
This study has relevance to clinical practice as it investigates patients' symptoms and their connection to biological markers and the subjective experience of general sickness.The hypothesis-driven approach with chosen biomarkers based on prior meta-analyses and experimental studies, contributes to the credibility and rigor of the study, reducing the risk for Type I and II errors.Some limitations should be taken into account when interpreting the results of this study.The observational design does not address questions of causality.The study's relatively small sample size could obscure associations, increasing the risk of type II errors or other methodological limitations.Another limitation is the lack of information about the different diagnoses of chronic pain.The participants suffer from mixed pain types, which adds some uncertainty to the interpretation of the results, although supporting the ecological validity as it represents the majority of chronic pain patients presenting at most pain clinics.We only have one time point for blood sampling, and it is known that variation in levels of cytokines in peripheral blood can be affected by factors such as time of day, physical activity, acute stress, sleep quality, and other factors (Nilsonne et al., 2016).This study was predominantly composed of women.Consequently, caution should be exercised when generalizing the findings to the male population or to a more heterogeneous sample.Nevertheless, the sample aligns with other studies involving participants with chronic pain in terms of the proportion of women, suggesting that the current sample is representative (Gerdle et al., 2019).

Future studies and clinical utility
Future studies with a larger sample and with a control group as a reference could use an adequate longitudinal design to investigate the interrelationship between the above-described symptoms and inflammatory markers in chronic pain to illuminate if these findings are different over time and in relation to patient characteristics.The temporal dynamics present a challenge in grasping the connection between inflammation and pain.Initially, inflammation often correlates with the onset of pain in various conditions.However, as time progresses, other pathological mechanisms may emerge as dominant factors, possibly undermining this correlation (Ren and Dubner, 2010).Thus, exploring this temporal dimension further is crucial for gaining a comprehensive understanding of the interplay between inflammation and pain.Inflammation can be beneficial for recovery in earlier phases of pain (Parisien et al., 2022), and in later phases, central inflammatory markers show complex relationships with pain severity.In chronic back pain disorders, several cerebrospinal fluid (CSF) inflammatory markers show a U-shaped relationship with pain severity (Rosenström et al., 2024), and in painful knee osteoarthritis, several CSF inflammatory markers show inverted relationships with pain severity (Palada et al., 2020).A larger sample could facilitate subgroup analyses, investigating if e.g., sickness behavior or psychiatric co-morbidity correlate with the level of cytokines for some individuals with chronic pain.Increasing the sample size and securing for gender balance would also fortify the reliability and extend the applicability of results.Also, broader inflammatory biomarker profiles are a level of analysis that could be further analyzed.Importantly, understanding more about the experienced sickness, as assessed by sickness behavior and notably prevalent in this specific group, may help improve interventions for these symptoms, as a potential treatment target.

Conclusions
The patients in the present sample experienced substantial comorbid problems, yet their associations with the included inflammatory biomarkers were either absent or weak.Also, despite high levels of selfreported sickness behavior, overall the inflammatory status remained within the normal range.Intriguingly, ratings of sickness behavior were more important in explaining ratings of depression and insomnia when compared to the influence of the inflammatory markers.Thus, the present results point to the complexity of chronic pain, and the difficulties of identifying biomarkers.

Fig. 1 .
Fig. 1.A histogram illustrating the distribution of the dataset.

Table 2
Clinical variables.

Table 3
Correlation matrix for study variables.

Table 4
Linear regression analyses with depression and insomnia as dependent variables.* Significant at the 0.001 level.* Significant at the 0.05 level.Adjusted R2 in regression model with depression as dependent variable = 0.49 Adjusted R2 in regression model with insomnia as dependent variable = 0.31 N = 71 CRP: C-reactive protein; IL-6: interleukin 6; TNF-α: tumor necrosis factor alpha. *