Pain is not simply determined by the intensity of the nociceptive stimulus but also by orchestrated mechanisms that work together, including psychological factors [1]. As one of these psychological factors, anxiety involves both physiological and psychological aspects that affect the way sensory interpretation occurs [2]. Anxiety is considered adaptive because a threatening situation induces body changes that increase the state of arousal. Anxiety can be presented as a state-anxiety (i.e., referred to acute situation-driven episodes that fluctuate over time) or as a trait-anxiety (i.e., a lifelong pattern, in the form of a personality feature) [3]. Maladaptive anxiety can take over and have negative effects on people’s lives [4].

These descriptions are consistent with the State-Trait Anxiety theory, which predicts that individuals with high trait-anxiety are generally hypersensitive to stimuli and are psychologically more reactive [3, 4]. Anxious patients present signs of restlessness, sympathetic overactivity, and resistance to sedation [5]. Thus, similarly to pain mechanisms, it is conceivable that anxiety is associated with alterations in brain excitability [5]. Previous reports observed that the motor-evoked potentials (MEP) amplitude was correlated with neuroticism in subjects with anxiety-related personality traits [6] and that short intracortical inhibition (SICI) was decreased in those with obsessive-compulsive disorder (OCD) when compared to subjects who were screened as psychiatrically normal [7].

In the clinical setting, subjects who frequently experience stress and anxiety have higher predisposition to develop trigger points, which can lead to myofascial pain syndrome (MPS) [8]. According to epidemiologic studies, the myofascial trigger points (MTrPs) might be a source of nociceptive inputs in 30% to 85% of the patients seeking pain relief [9, 10]. A common sign of MPS is the referred pain to distant somatic structures with concomitant modifications of superficial and deep sensitivity in the painful areas [11–13]. It has been theorized that there are two main mechanisms for pain hypersensitivity in muscle pain syndromes: a “bottom-up mechanism” in which the process starts and is maintained by changes in deep peripheral tissues, and the “top-down mechanism” which claims that the origin lies in the stress- and pain- regulating systems in the brain [14]. Both mechanisms would induce changes in the excitability of nociceptive neurons in the dorsal horn, descending and ascending corticospinal tracts and cortical structures, leading to hypersensitization. To study the inhibitory mechanisms that modulate pain processing at the spinal cord, a “conditioned pain modulation (CPM) stimulus” [15] method has been used in which heterotopic noxious stimuli applied to a remote area of the body was shown to attenuate the activity of pain-signaling neurons in the spinal dorsal horn [16–18].

To improve the understanding of the central mechanisms related to anxiety and pain, we assessed cortical excitability parameters using single and paired pulse transcranial magnetic stimulation (TMS). We hypothesize that corticospinal excitability is modulated by anxiety favoring loss of descendent inhibitory influx. The study presented here aimed to answer three questions: 1) Is the motor cortex excitability, as assessed by transcranial magnetic stimulation parameters (TMS), related to state-trait anxiety? 2) Does anxiety modulate corticospinal excitability changes after evoked pain by Quantitative Sensory Testing (QST)? 3) Does the state-trait anxiety predict the response to pain evoked by QST if simultaneously receiving a heterotopic stimulus [Conditional Pain Modulation (CPM)]?

We screened 62 potential participants with a diagnosis of MPS, and forty-seven of them were included in the study. The reasons for exclusion were not fulfilling diagnostic criteria for MPS, lack of disability as defined in the protocol, inability to demonstrate a neuropathic component as assessed by DN4 (Neuropathic Pain Diagnostic Questionnaire) or the presence of another diagnosis (fibromyalgia).

In Figure 2 are presented the MEP amplitude in mV, before and after QST of a representative patient. This data presented is an average of 10 trials before and after QST. Patients presented higher MEPs amplitudes after QST in relation to the amplitudes before.

Eleven healthy controls with mean (standard deviation) age of 26.27 (6.31) were recruited. Their mean trait-anxiety score was 17.18 (3.43), and state-anxiety score 21.55 (4.72). Their MEP remains unchanged before and after the QST (P > 0.05) (Figure 3).

Effects of analgesics on the relationship between anxiety and the corticospinal modulatory system

The relationship between the percentage in pain reduction in NPS with trait-anxiety and DRP is presented in Figure 4. It was observed an inverse correlation between NPS during CPM-TASK in both DRP and trait-anxiety. This association between CPM, trait-anxiety and DRP observed in this univariate analysis was confirmed using multiple linear regressions (Table 5). This multivariate analysis confirmed that there is a significant inverse correlation between NPS during CPM-TASK and DRP, whereas the trait-anxiety was marginally significant (P = 0.05, see Table 5). A significant interaction between trait-anxiety and DRP was detected, when adjusting for this interaction one unit increment on this product (trait-anxiety vs. DRP) was associated with 39% increment on the NRS (P = 0.02) (Table 5). That is, the efficiency of the corticospinal system was reduced for modulating the pain evoked by QST. Also, this model showed that analgesics use did not change the relationship between anxiety levels and corticospinal excitability (P > 0.05).

Source	Type III sum of squares	df	Mean square error	F	P
Corrected model	24327.85	4	26219.170	2.05	0.10
Dependent variable: Percentage of change in the pain score on numerical pain scale during CPM-TASK a
Parameter	β	SEM	T	P	Partial eta squared
Intercept	458.80	221.72	2.07	0.04	0.02
B-PCP:Sb	−10.56	3.82	−2.76	0.01	0.15
Trait-anxiety	−21.55	10.72	−2.01	0.05	0.09
B-PCP:S* Trait-anxiety	0.39	0.17	2.38	0.02	0.12
Analgesic dose used weeklyc	−0.03	0.05	−0.07	0.94	0.00

^a Conditional pain modulation (CPM).

^b Profile of Chronic Pain: Screen for a Brazilian population (B-PCP:S).

^b Number of days analgesics were used per week in the last three months (< 4 times/≥ 4 times).

This study suggests that there is a relationship between motor cortex excitability and trait-anxiety. We found a positive correlation between trait-anxiety, the DRP and the ICF. Trait-anxiety was inversely correlated with the CSP. Additionally, we observed that when high trait-anxiety concurs with DRP, the effect of the heterotopic noxious stimulus on the pain induced during the QST was reduced (Table 5). High trait-anxiety is positively correlated with increased cortical excitability and decreased intracortical inhibition as assessed by ICF and CSP (Table 4).

According to previous reports, the inhibitory deficits mediated by GABAergic receptors associated with manifestations of anxious behavior are coherent with our findings [6, 34, 35]. A marked increase in ICF has been demonstrated in drug-naive subjects suffering post-traumatic stress disorder (PTSD) [36], as well as a short CSP in patients with obsessive-compulsive disorder [7]. Although the underlying mechanism is not clear, several neurobiological processes may explain these findings, as follows. The ICF originates from excitatory postsynaptic potentials (PSPs) mainly mediated by glutamatargic NMDA receptors [37]. Their latency is about 10 ms, consistent with the time course of intracortical facilitation [38]. Pharmacological studies support such observation, as NMDA receptor antagonists (i.e. dextromethorphan) decrease ICF [38]. Likewise, GABA-A agonists like benzodiazepines (e.g. Lorazepam) and barbiturates decrease the SICF. This supports the hypothesis that the first pulse elicits the GABA-A receptor-mediated short latency inhibitory postsynaptic potential (PSPs) in corticospinal and/or first order excitatory interneurons that inhibits the facilitatory interactions with the second pulse [39]. On the other hand, GABA-B agonists (e.g. Baclofen) increase ICF [40]. It has also been suggested that ICF is not exclusively mediated by excitatory interneurons, but rather by a balance between inhibition and excitation [38].

Finally, the duration of the cortical inhibition assessed by CSP is consistent with intracellular measurements of the inhibitory PSPs from the stimulation of the GABA-A receptor [41, 42]. Also, a short-lasting CSP shortening after diazepam injection has been reported [43]. It has been hypothesized that, just after diazepam injection, a GABA-A mediated inhibition could transiently reduce facilitatory thalamo-cortical influences on inhibitory interneurons of the motor cortex [43]. Overall, the changes in the balance between cortical inhibitory and facilitation observed in the present study may be explained by impairments in neurotransmission mediated by GABA-B and NMDA receptors [44].

We found that trait-anxiety explained about 36% of the variance of the MEP amplitude induced by QST evoked pain. Patients with high trait-anxiety might have less corticospinal modulation of the pain response during QST, which also made them present higher motor cortex excitability. Furthermore, it is unlikely that time or other non-specific effects of the QST intervention might have influenced the cortical excitability parameters, because healthy controls MEPs remained unchanged after the QST intervention. This finding is in agreement with previous reports that showed a short latency during a stimulus on peripheral sensory nerves induced by QST in anxious subjects [45]. Previous reports had consistently described increments on MEP after painful experiences [46–48] and under experimental pain [48–51]. To the best of our knowledge, the present study is the first assessment of the association between anxiety and MEP changes in patients with chronic pain submitted to standardized nociceptive stimuli. Thereby this finding suggests that the inhibitory capacity of the corticospinal modulator system is reduced. In fact, this effect in chronic pain patients underscores dysfunctional cortical processing in these subjects, because the motor reactions to pain in healthy controls result in the suppression of MEP amplitude from all distal hand muscles [48–51]. Another potential explanation to be considered is the potential protective effect of emotional amplification on pain consistent with the Gray–McNaughton theory, which proposes that during anxiety the hippocampal formation increases the valence of aversive events to prime behavioral responses adaptive to the worst possible outcome. It has been described that the hippocampal formation is responsible for increased pain by amplifying signals to the neural representation of the painful stimulus [52]. In fact this response may change overall processing of pain on the inhibitory system as compared to experimental pain (or acute pain) in healthy subjects. Overall, our observations help to interpret the electrophysiological evidence for the relationship between psychological symptoms and pain modulation.

Furthermore, it was also observed that the interaction between pain and anxiety reduced the CPM during the QST evoked pain. Considering that the CPM evaluates the function of the corticospinal system, a lower pain threshold was observed when the high trait-anxiety was concurrently with high DRP, this suggest that the function of the descending modulatory system was reduced. Such a finding is in agreement with previous reports in other chronic pain conditions, such as temporomandibular disorder [53]; fibromyalgia [54, 55]; tension-type headache [56]; migraine [57] and also in healthy subjects with significant pain history [58]. Overall, these findings indicate that the impact of trait-anxiety on pain is linked with the central sensitization of nociceptive neurons, which contribute to the worsening of chronic pain symptoms. Nevertheless, when assessing the association between trait-anxiety and Diffuse Noxious Inhibitory Control (DNIC) induced by the immersion of one’s hand at 12°C [58, 59]. Edwards et al did not find an association between the extent of endogenous analgesia (tested by DNIC) and psychological parameters, including the profile of mood states, locus of control, level of vigilance, and stress. Such divergence between our findings and those of Edwards et al.[58, 59] could be explained by differences in the samples (such as the inclusion of healthy male subjects), the intensity of the stimuli used (12°C vs. 0°C) and type of assessment conducted.

We observed not only evidence of increased central sensitization in chronic pain associated with anxiety but also a negative impact of trait-anxiety on the endogenous modulatory system in patients with long-term pain associated with higher levels of disability. Thus, a potential mechanism to explain such effects may be related to a change in the neural response from amygdala. Healthy subjects’ pain modulation is attributed to endogenous opioid activity activated by the amygdala [60]. The amygdala is directly connected to brainstem structures responsible for the descending modulation, which is at the same time regulated by endogenous opioid activity [60]. Thus, in patients with chronic pain, although greater activation of the amygdale might occur, a reduced opioid activity is expected because of the central sensitization induced by the chronic condition, leading to an attenuated endogenous analgesic response [61]. Such hypothesized explanatory mechanisms for our observations presented here might make us consider the endogenous modulation system as a potential target for specific approaches in chronic pain control in the context of higher anxiety levels.

This study has some limitations. First, the use of TMS for neurophysiological assessments involved the evaluation of neurotransmitter system activity in an indirect manner, and it has been shown to have relatively low specificity. However, TMS provides a useful tool for neurophysiological assessment because it induces activity and evaluates a subject’s response. Second, we included only female subjects, although an enhanced pain response in females has been attributed to physiological and psychological variables, including mechanisms of endogenous inhibition, the capability to endure pain, genetic factors, pain expectation and personality traits [62, 63]. In this context, the gender may be an important confounding factor because the amygdala is more prone to activation upon negative emotional responses (i.e., stress, fear, and anxiety) in females [64]. Another factor to consider is the hormonal variation throughout the menstrual cycle. Although this factor is a possible confounder, the effect of estrogen levels on the MEP after QST evoked pain can be minimized because each patient served as her own control. Third, another limitation to be considered is the way our team evaluated the use of analgesics because it is always possible to face a memory bias when inquiring patients about past events. It is possible that those experiencing more pain, or those with higher anxiety traits, would report a higher requirement of analgesics than those who were less anxious or experiencing less pain. Fourth, although we recruited healthy volunteers to assess the potential effect of time during the experiment, it is worth noting that our control sample was younger on average. Nevertheless, age only showed an effect on the motor threshold in the regression models, but not in the other cortical excitability parameters. Thus, it is unlikely that controls’ age would modify the present findings. Finally, it would be worthwhile to design similar studies to investigate anxiety’s effect on other chronic pain syndromes to confirm the findings revealed in the present study because our conclusions can only be claimed as valid in similar severe chronic MPS samples. Fifth, chronic pain is strongly associated with anxiety symptoms; this potential confounding factor cannot be fully controlled, as it is part of chronic pain syndromes [65, 66]. It is important to emphasize that in the single-subject design the patients serve as own control. This design that is sensitive to individual organism’s differences permits to assess causal relations between the independent and dependent variables [67, 68]. Whereas it reduces the potential of comparing with healthy subjects, it is an ideal strategy to validate results because in real life a scenario is complex to find controls that match with the profile of the chronic pain patients.

In conclusion, we showed that when the chronic MPS concurs with high trait-anxiety and higher DRP pain, an imbalance occurs between excitatory and inhibitory impulses in the descending systems to the dorsal horn (as indexed by ICF, CSP and MEP change upon pain); thus providing further support to the effects of the emotional system on central sensitization and also to the potential mechanisms of decreased pain modulation in chronic pain.