Conditioned pain modulation (CPM) and temporal summation (TS) tests can measure the ability to inhibit pain in fibromyalgia syndrome (FMS) patients and its level of pain sensitization, respectively. However, their clinical validity is still unclear. We studied the association between changes in the CPM and TS tests and the clinical improvement of FMS patients who received therapeutic intervention. We systematically searched for FMS randomized clinical trials with data on therapeutic interventions comparing clinical improvement (pain intensity and symptom severity reduction), CPM, and TS changes relative to control interventions. To study the relationship between TS/CPM and clinical measures, we performed a meta-regression analysis to calculate odds ratios. We included nine studies (484 participants). We found no significant changes in TS or CPM by studying all the interventions together. Our findings show that this lack of difference is likely because pharmacological and non-pharmacological interventions resulted in contrary effects. Non-pharmacological interventions, such as non-invasive neuromodulation, showed the largest effects normalizing CPM/TS. Meta-regression was significantly associated with pain reduction and symptom severity improvement with normalization of TS and CPM. We demonstrate an association between clinical improvement and TS/CPM normalization in FMS patients. Thus, the TS and CPM tests could be surrogate biomarkers in FMS management. Recovering defective endogenous pain modulation mechanisms by targeted non-pharmacological interventions may help establish long-term clinical recovery in FMS patients.
Fibromyalgia syndrome (FMS) is a chronic rheumatic disease characterized by diverse symptoms such as widespread body pain, stiffness, fatigue, sleep disturbances, cognitive impairments, and psychiatric symptoms, which significantly impact the patient’s daily routine and quality of life. Furthermore, it often requires healthcare services, becoming a costly public health issue.
Central sensitization consists of a variety of neural system dysfunctions, and it provides a possible explanation for comprehension and management of chronic pain syndromes such as fibromyalgia. This theory suggests that abnormal sensory processing in the brain, increased susceptibility to stimulus, and inaccurate pain modulation pathways result in hypersensitivity to stimulus. In addition, Mezhov et al. (2021) argue that the presence of central sensitivity syndromes, such as fibromyalgia, may reflect an overactive central nervous system instead of an isolated disorder, associated with changes in brain oscillatory activity. Recognizing central sensitization as a mechanism in chronic pain increases the importance of treatments focusing on modulating the endogenous pain inhibitory system instead of short-term treatment that only provides temporary pain relief by blocking nociceptive input. This change in pain management provides awareness and highlights the importance of targeting pain compensatory mechanisms as a new approach to manage FMS.
Conditioned pain modulation (CPM) and pain temporal summation (TS) are key to understanding pain modulation dynamics and can assess the pain sensitization and pain inhibition characteristics of FMS patients non-invasively; they are also promising biomarkers. CPM involves the inhibition of one painful stimulus by another, reflecting the body’s endogenous pain control systems and assessing their efficiency in both healthy and chronic-pain populations. TS describes the increased perception of pain from repeated noxious stimuli, highlighting excitatory processes in central pain facilitation. However, there is limited evidence supporting the validity of these biomarkers, mainly due to contradictory results associating CPM metrics with clinical outcomes. O’Brien et al. (2018) found that FM patients show a heightened response to repetitive noxious stimuli and reduced CPM compared to healthy controls, supporting the idea of central sensitization and defective pain inhibition. Similarly, Potvin and Marchand (2016) reported that CPM procedure has a specificity of 78.9%, but its low sensitivity of 45.7% indicates that it may not be detected in all patients. These findings suggest that endogenous pain inhibition mechanisms are impaired in these patients, but only partially, which indicates the complexity and unpredictability of pain modulation in this condition. Conversely, a study found no correlation between pain severity and CPM efficacy, challenging the idea that CPM is a reliable biomarker. Similarly, a different study reported no significant difference in CPM between FM patients and those with generalized pain, indicating diverse pain phenotypes and mechanisms in FM syndrome. The limited sample size and the cross-sectional nature of the analysis, not considering the dynamic features of endogenous pain modulation, indicate the presence of limitations in previous studies. To our knowledge, no previous studies tried to validate systematically the longitudinal changes in CPM/TS with changes in clinical outcomes such as pain intensity or symptoms severity in FMS. A longitudinal validation of these biomarkers is needed to foster the utilization of objective metrics in the context of FMS management and the development of personalized therapeutic protocols.
Therefore, we aim to explore the association between changes of endogenous pain modulation (indexed by CPM and TS) and clinical changes (pain intensity and symptoms severity) after therapeutic interventions in FMS patients. We hypothesize that improvement in endogenous pain modulation will correlate with the improvement of clinical profiles in FMS patients.
This systematic review and meta-analysis were conducted following the “Preferred Reporting Items for Systematic Reviews and Meta-Analyzes” (PRISMA) guidelines and the Cochrane Handbook for Systematic Reviews of Interventions. The protocol was registered from the Open Science Framework platform with the code YDG4J.
PubMed/MEDLINE and Embase databases were searched from inception until 4 March 2024. The complete search strategy was: (“Fibromyalgia” [MeSH] OR “Fibromyalgia” [TIAB]) AND (“Sensory profile” OR “temporal summation” OR “conditioned pain modulation” OR “temporal slow pain summation” OR “quantitative sensory testing” OR “cognitive-emotional sensitization” OR “sensory threshold” OR “pain threshold” OR “diffuse noxious inhibitory control” OR “heterotopic noxious conditioning stimulation” OR “endogenous analgesia” OR “pain inhibition” OR “endogenous pain modulation”). Additionally, we reviewed the references of the included studies. The eligibility criteria were: (1) Randomized controlled trials (RCTs) that included FM patients and assessed CPM/TS before and after an intervention and that included at least one clinical pain measure (e.g., VAS, FIQR); (2) any type of intervention and comparison; and (3) full-text accessible. The exclusion criteria were: (1) pre-clinical studies; (2) review articles; (3) letters to the editor and editorials; and (4) conference abstracts. Studies were not excluded based on date or language.
Duplicate records were removed manually using the Covidence web platform. Two independent reviewers conducted the screening, first by titles and abstracts and then by full text, with the aid of the Covidence web platform. Disagreements were solved by the facilitation of a third author. The data extraction was conducted by two authors independently. The following variables of interest were extracted: study design, sample size, CPM/TS method, type of stimulus, type of control group, intervention type, fibromyalgia diagnosis description, age, gender, pain intensity, and symptom severity scores indexed by the Revised Fibromyalgia Impact Questionnaire (FIQR).
Unavailable raw data from the main outcome were calculated from available graphs using Web Plot Digitizer v.3.11 when possible. In the cases where data from graphs were not available, authors were contacted via email. If no response was received by the time of the analysis, the study was excluded.
Assessment of the studies’ risk of bias was carried by two independent reviewers, and discrepancies were solved with the support of a third author. We used the Cochrane Risk of Bias 2 (RoB 2) tool. We used the algorithm proposed in the RoB 2 tool to rate each domain as ‘low risk of bias’, ‘some concerns’, or ‘high risk of bias’, with the overall risk of bias being the worst assessment of the five domains. We did not conduct a publication bias assessment due to the small number of included studies.
We conducted meta-analyses of continuous outcomes using random-effects models due to the high expected heterogeneity. The DerSimonian–Laird method was used. Each variable (visual analogue scale (VAS), FIQR, CPM, and TS) was analyzed separately. Additionally, we converted the CPM and TS differences as percentage change as sensitivity analysis. Then, we calculated effect sizes as standardized mean differences (MD) with 95% confidence intervals (CIs).
Moreover, we used the Hartung–Knapp adjustment for random effects models, which calculates more adequate error rates, especially when the number of included studies is small. We tested for statistical heterogeneity of pooled estimates using the Chi 2 test and the degree of heterogeneity using the I 2 statistic, considering that heterogeneity might not be substantial when I 2< 40%. We did not perform a publication bias assessment due to the low number of studies we found.
For the meta-regression analyses, we categorized each improvement in CPM and TS as an increase of at least 10% from the baseline in each trial arm (active and control). Similarly, we categorized the studies as clinical improvement based on an improvement on at least 20% of the scale from baseline, based on the minimal clinically important difference (MCID) for the FIQR score and the VAS score. Then, we constructed two-by-two tables to calculate individual odds ratios (OR) and corresponding 95% CI using the Woolf approximation. The data were processed with R Studio 4.1.1 for the creation of the plots (R Foundation for Statistical Computing, Vienna, Austria).
Our literature search retrieved 1598 records. Of those, nine were included after full-text assessment (484 participants). The flowchart of the selection process is presented in Figure 1.
The interventions used were neuromodulation (n = 3), pharmacological (n = 5), and education (n = 1). Among these nine studies, four exclusively utilized CPM, one used only TS, and four incorporated both paradigms. Mechanical modalities (pain pressure threshold (PPT)) emerged as the predominant choice for conditioned stimulation (n = 7), while heat sensors were utilized in the remaining studies (n = 2). Concerning the conditioning stimulation in CPM, the cold water was the most commonly used (n = 4), with occlusion cuffs (n = 3) and hot water (n = 1) also employed in other studies. A detailed description of the included studies is found in Table 1.
Overall, the studies demonstrated appropriate randomization procedures and adhered closely to the intended interventions, contributing to an overall assessment of low bias risk. However, one study presented some concerns regarding missing outcome data, primarily due to a significant number of participants in the intervention group experiencing collateral effects, which resulted in a high withdrawal rate from the study. Another study was judged to have some concerns regarding deviations from intended interventions in certain studies, primarily due to the absence of blinding and the lack of a sham intervention. These factors potentially introduced biases, as they could influence the result of the outcome. The full assessment is presented in Table 2.
Based on the data from the included studies, VAS scores were analyzed to compare the effects of the experimental intervention against the control group, showing a decrease in VAS scores in intervention group. However, the pooled effect size indicated a reduction in VAS scores with a wide confidence interval that suggests the overall effect is not statistically significant, SMD = 1.15, 95% CI [−3.70; 1.40]. The heterogeneity test revealed significant variability (I 2 = 95%, p< 0.01), meaning that the intervention effects were inconsistent across different studies (Figure 2A).
The analysis of TS demonstrated varied effects between the experimental and control groups. The pooled effect size SMD = 0.48, 95% CI [−2.96; 3.92], indicated an increase in TS scores with a wide confidence interval not significant. The test of heterogeneity also revealed
