Probably the most frequently reported neuroimaging studies comparing people with chronic pain and healthy controls is structural MRI, gray matter differences (VBM) or CTA. While the brain regions involved are somewhat variable, overwhelmingly, the results of these many studies indicate gray matter density loss in multiple brain areas is a common consequence of chronic pain (see reviews by May (2008) and Schweinhardt et al. (2008)); in particular, these studies implicate the cingulate and insular cortices, as well as prefrontal and parietal regions. Studies have also reported that decreased gray matter density or cortical thickness in patients is reversible with effective treatment (Gwilym et al. 2010; Rodriguez-Raecke et al. 2009; Seminowicz et al. 2010). A longitudinal study in a rat model of neuropathic pain reported prefrontal cortical volume decreases several months after the injury, and this morphometric change coincided with the onset of anxiety-like behaviors (Seminowicz et al. 2009). Thus, the evidence indicates that chronic pain causes brain atrophy, and that relief from chronic pain can restore normal brain anatomy. Furthermore, affective changes associated with chronic pain are likely partly responsible for these brain changes.

A few studies to date have examined the structural connectivity changes with DTI. Geha et al. (2008a) reported aberrant connectivity patterns from the anterior insular cortex in people with chronic regional pain syndrome compared to healthy controls, as mentioned in section “Diffusion Tensor Imaging.” Other groups have shown altered white matter tracts in migraine (Moulton et al. 2011), CPSP (Seghier et al. 2005), and fibromyalgia (Lutz et al. 2008; Sundgren et al. 2007). This technique seems particularly helpful when used in combination with fMRI or VBM, where one can show a relationship between abnormal structure or function, and connections to or from that region.

In resting state fMRI, subjects are usually asked to lie awake and relaxed in the scanner, with eyes open or closed. This technique is useful for assessing functional networks in the resting brain, i.e., without any task imposed. Several resting state networks have been identified, including a sensorimotor network (Biswal et al. 1995), a default mode network (DMN; Greicius et al. 2003; Raichle et al. 2001), which is a set of regions normally deactivated during task execution (i.e., areas that are active at rest), and several others (De Luca et al. 2005).

Two anti-correlate networks were identified by Fox et al. (2005), and called the task-positive network, which resembles the set of regions normally activated by performance of a cognitively demanding task, and a task-negative network, which resembles the DMN. Seminowicz and Davis (2007) reported that this task-positive network could be enhanced in healthy individuals when pain was evoked with electrical nerve stimulation, suggesting that pain acts as an additional cognitive load (Fig. 27.3). Patients with chronic low back pain have reduced deactivations in task-negative areas (Baliki et al. 2008), suggesting that these areas might be less active at rest, perhaps because of hyperactivity of task-positive regions. Mantini et al. (2009) supported these findings, reporting that in healthy individuals pain evoked with electrical nerve stimulation led to reduced DMN activity. Thus, chronic and experimental pain can modulate brain activity at rest, and this effect might be explained by the effects of pain on cognitive networks.

Other studies have reported altered resting state activity in chronic pain conditions, including changes in spatial networks (Cauda et al. 2009, 2010; Napadow et al. 2010; Tagliazucchi et al. 2010), and in the temporal (frequency) aspects (Malinen et al. 2010). One of the great advantages of resting state fMRI is that many different networks – cognitive, salience (Seeley et al. 2007; Taylor et al. 2009), or sensorimotor – can be interrogated from a single dataset. Data sharing projects now let researchers access tens of thousands of resting state scans from different settings, and this will allow for comparisons of resting state networks in many different disorders.

Although cognitive behavioral therapy (CBT) interventions have been used in the management of persistent pain for some time (see also Donovan et al. 2011), relatively little attention has been given to the mechanisms by which CBT achieves pain relief. To our knowledge, only one published study has examined the effects of a CBT protocol on patterns of brain activation during a painful inducing task (rectal balloon distension) (Lackner et al. 2006). In this study, the responses of six irritable bowel syndrome patients who had undergone CBT were compared to that of five healthy controls. All patients in the study underwent an imaging session before and after completing a 10-week CBT protocol which included training in several coping skills such as self-monitoring, cognitive reappraisal of pain, strategies for controlling worry, and problem solving. Data analyses showed that, after completing CBT, patients had reductions in neural activity in the parahippocampal gyrus and inferior portion of the right cingulate cortex when compared to both their pretreatment responses and the responses of healthy controls. Interestingly, the magnitude of change in these neural responses was found to be significantly correlated with reductions in a number of gastrointestinal symptoms including pain, as well as reductions in anxiety. The authors concluded that changes in the activation of brain structures, involved in attention and emotion regulation, may represent one mechanism by which CBT achieves its effects on pain and other symptoms of irritable bowel syndrome. Naylor et al. (manuscripts submitted for publication) studied the effects of CBT on brain function and structure in patients with chronic musculoskeletal pain. After 11 weeks of CBT patients showed increased activity in the insula, reduced activity in the amygdala and frontal areas associated with emotional processes of pain. This activity was correlated with subjective measures of pain and coping. In this study Naylor et al. also assessed the neuroplasticity of gray matter atrophy previously reported in patients with chronic musculoskeletal pain. CBT was associated with increased gray matter volume (GMV) in DLPFC and aACC. The post-CBT increase in gray matter volume was correlated with a decrease of pain catastrophizing (Naylor et al).

Given the interest in the neural correlates of CBT treatments for emotional disorders (e.g., panic disorder) (de Carvalho et al. 2010), the relative lack of studies of CBT for pain relief is surprising. This is clearly an important direction for future research.

Interest in the neural correlates of placebo analgesia has been high (see also Pollo and Benedetti 2011). A number of studies support the notion that placebo analgesia is related to activation of the brain regions involved in endogenous opioid pain modulation (Bingel et al. 2006; Petrovic et al. 2010; Wager et al. 2004; Zubieta et al. 2005) – the same regions involved in opioid analgesia.

An interesting possibility is that placebo analgesia not only affects pain processing at the level of the brain but also at the level of the spinal cord. Eippert et al. (2009) used fMRI imaging of the cervical spinal cord to test this possibility. Administration of a placebo agent, which reduced ratings of thermal pain by 26%, what posterior parietal cortex, left DLPFC found to be associated with a ­significant reduction in BOLD responses at the ipsilateral dorsal horn of the spinal cord. These results are consistent with the notion that descending pain control systems responsible for placebo analgesia influence the gating of nociceptive input at the level of the spinal cord.

Recently, researchers have been interested in the effects of placebo conditioning trials on the effects of placebo analgesia. In placebo conditioning trials, patients are provided with repeated exposures to a placebo (e.g., a placebo cream applied to an area of pain stimulation) and this is matched with intentionally lowered level of pain stimulation to increase patients’ expectations of placebo analgesia. Watson et al. (2009) studied the effects of placebo conditioning on brain activity occurring in response to a laser pain stimulus. Participants, 11 healthy individuals, underwent brain imaging before, during, and after the placebo conditioning trials as well as administration of the placebo analgesia. Data were collected during both anticipation of pain stimulation and during pain stimulation. Anticipation of pain stimulation was associated with activation of the left DLPFC, medial frontal cortex, and anterior mid-cingulate cortex for both the placebo conditioning and placebo administration conditions. Delivery of pain stimulation was associated with activation of the anterior MCC and posterior cingulate. The authors suggest that their findings support the notion that placebo conditioning produces activations in prefrontal and mid-cingulate cortices which persist during placebo analgesia.

Investigators have examined the possibility that there may be differences in the mechanisms underpinning opioid analgesia and endogenous opioid analgesia. Petrovic et al. (2010) argued that during administration of opioids, patients’ expectations that they will obtain effective pain relief are met and are matched with corresponding changes in the processing of nociceptive input. This contrasts with a placebo condition during which patients’ expectations for effective pain relief are not necessarily matched by corresponding changes in the processing of nociceptive input. Given prior findings that prefrontal mechanisms are involved in expectation processes and the matching of expectations with sensory input, Petrovic et al. (2010) hypothesized that these regions might be more active in placebo analgesia vs. opioid analgesia. To test this possibility, they reanalyzed data from an fMRI study contrasting opioid analgesia and placebo analgesia. Their findings indicated that, when compared to opioid analgesia, placebo analgesia was associated with substantial increases in activity in prefrontal areas (specifically, the lateral OFC and ventrolateral PFC). A connectivity analysis showed that activation of these prefrontal regions covaried with activation of the right ACC. The authors concluded that the mechanism of placebo analgesia is distinct from opioid analgesia in that it involves greater activation in prefrontal brain areas related to the processing of how expectations fit with sensory input.