Cerebral Perfusion in Patients with Chronic Neck and Upper Back Pain: Preliminary Observations
Cerebral Perfusion in Patients with Chronic Neck and Upper Back Pain: Preliminary Observations
Maxim A. Bakhtadze, MD, PhD, Howard Vernon, DC, PhD, Anatoliy V. Karalkin, MD, PhD, Sergey P. Pasha, MD, PhD, Igor O. Tomashevskiy, MD, PhD, David Soave, MSc
OBJECTIVE: The purpose of this study was to determine the correlation between cerebral perfusion levels, Neck Disability Index (NDI) scores, and spinal joint fixations in patients with neck pain.
METHODS: Forty-five adult patients (29 were female) with chronic neck/upper thoracic pain during exacerbation were studied. The subjects were grouped according to NDI scores: mild, moderate, and severe. The number of painful/blocked segments in the cervical and upper thoracic spine and costovertebral joints, pain intensity using the visual analog scale, and regional cerebral blood flow of the brain using single-photon emission computed tomography (SPECT) were obtained. The SPECT was analyzed semi-quantitatively. Analysis of variance tests were conducted on total SPECT scores in each of the NDI groups (P < .05). Univariate correlations were obtained between blockage, pain, and SPECT scores, as well as age and duration. A multivariate analysis was then conducted.
RESULTS: Group 1 (mild) consisted of 14 patients. Cerebral perfusion measured by SPECT was normal in all 8 brain regions.
Group 2 (moderate) consisted of 16 patients. In this group, a decrease in cerebral perfusion was observed (range, 20%-35%), predominantly in the parietal and frontal zones.
Group 3 (severe) consisted of 15 patients. In this group, the decrease in cerebral perfusion observed was from 30% to 45%, again predominantly in the parietal and frontal zones.
A significant difference was found between NDI groups (“moderate” and “severe” showed significantly greater hypoperfusion than “mild”). Total blockage score correlated with SPECT scores at r = 0.47, P = .001. In a multivariate analysis, NDI scores contributed 39% of the variance of SPECT scores.
CONCLUSION: In this group of patients with neck and/or upper back pain, NDI scores strongly predicted cerebral hypoperfusion. Spinal joint dysfunction may be involved via hyperactivity in the regional sympathetic nervous system.
From the Full-Text Article:
The primary goal of this study was to determine if categorizing patients with chronic neck pain by their NDI scores could distinguish those with and without cerebral perfusion abnormalities, thereby improving the efficiency of using SPECT for this purpose. Our findings strongly support this result. Single-photon emission computed tomography scores in NDI groups 2 and 3 were significantly higher than in group 1. Neck Disability Index scores strongly correlated with SPECT scores, and NDI scores were the only variable loading onto SPECT scores, contributing 39% of the variance of those scores.
Our findings replicate the strong correlation between NDI and PET scores reported by Linnman et al,  with their finding of temporo-occipital hypoperfusion in patients with WAD. This correlation between disability level and perfusion status may explain some of the inconsistencies in the previous studies of SPECT in neck pain. [23-31] According to our finding, it is the severity of self-reported disability that appears to be associated with cerebral perfusion abnormalities, not the simpler status of WAD vs non-WAD.
The location of reduced rCBF findings in our study—parietal and frontal lobes in groups 2 and 3 — is not in accord with those reported for non-WAD cases by Sundstrom et al  and Guez ; that is, hypoperfusion in the right temporal region. Our findings are somewhat more similar to those reported by Otte et al [23-26] for patients with WAD—parieto-occipital hypoperfusion. However, technical and methodological issues (see “Limitations” below) make comparisons between these studies difficult.
An explanation for the association between NDI and SPECT scores requires discussion of at least 3 separate domains. First, there is now strong evidence that chronic pain itself is associated with changes in the brain. [46-49] Although we did find that pain level does correlate with SPECT scores, our finding that it is the severity of self-reported disability that statistically distinguishes those with and without such changes is a novel finding, at least for patients with neck pain.
Second, there is now strong evidence of an association between NDI scores and psychosocial variables in patients with neck pain. Several studies have now identified strong correlations between NDI scores and fear-avoidance beliefs, [50-58] catastrophizing, [59, 60] depression, [58, 61] and anxiety.58 There is also strong evidence for altered brain processes in chronic depression. [62-64]
This creates a nexus of factors: self-rated disability, psychological decompensation and brain changes, which are very challenging to decode. At this point, it is not possible to determine the causal direction of these associations within these first 2 domains, leading to more questions than answers: does chronic pain change the brain and lead to psychological decompensation? or do brain changes associated with depression and other manifestations of anxiety promote pain chronicity? What role does the variable “self-rated disability” (NDI scores) play in this nexus? There is currently great interest in applying the Fear-Avoidance Behavior Model to whiplash, specifically, and chronic neck pain in general. [50-54] Contributions to this model from each of the factors noted previously are just emerging in the literature. 
A third domain of consideration is a set of physical findings that have also been strongly correlated with NDI scores, namely, reduced pain thresholds (tenderness), as measured by pressure pain and by heat pain, and changes in motor control patterns, all of which have been reported by Sterling and Pedler [20, 21] and Jull et al.  These changes have been suggested to result from central sensitization promoted by chronicity and severity of neck pain (at least in patients with WAD).
This theory that chronic cervical spinal pain produces nociceptively driven changes in somatosensory and sensory-motor changes (as well as somatoautonomic changes [see below]) was first advanced by Curatolo et al , Banic et al,  and Herren-Gerber et al  for neck pain. However, it actually has its origins in theories of referred pain and spinal hyperexcitability that arose in the early 20th century and were crystallized in the mid-20th century by Korr [70-73] and Denslow et al.  Based on that comprehensive theory (“the central excitatory state”), spinal pain was thought to induce alterations in somatoautonomic reflexes. The original idea (in modern terms) that central sensitization resulting from chronic spinal pain includes a domain of clinical effects based on disturbed segmentally organized sympathetic outflow remains elusive to investigation and clinical acceptance. Tantalizing evidence of this has been presented by Sato et al, [75, 76] Sato and Swenson,  Budgell et al, [78-83] Fujimoto et al,  and Bolton et al. [85-87]
The idea that sympathetic activation from spinal pain or stimulation of spinal sympathetic ganglia can be associated with or even cause cerebral hypoperfusion also has a long history (posterior cervical sympathetic syndrome of Barré, [88, 89] cervical migraine, and vertebrogenic migraine). [90-92] Nociceptive stimuli from dysfunctional motion segments of the spine activate the segmental sympathetic nervous system, and these somatic afferent stimuli will cause reactions in the preganglionic sympathetic neurons both segmentally [77-81, 85] and suprasegmentally. [75, 76]
Sympathetic ganglia will also respond to these somatic afferent stimuli. McLachlan et al  and colleagues registered synaptic events in neurons of the upper cervical sympathetic ganglion in response to noxious cutaneous stimuli in rats. Lichtman and colleagues  studied the patterns of end-organ sympathetic reaction in the territory supplied by the superior cervical sympathetic ganglion in response to stimulation of the thoracic ventral roots (predominantly Th1-Th5). One of their findings was constriction of the arterioles of the eye and ear. In 1980, the same investigators studied the territory of the stellate ganglion (SG), which also receives afferents from the Th1 to Th5 spinal segments.  Their work gave a fuller understanding of how irritation from the spinal segments Th1 to Th5 may result in such a wide sympathetic response from end-organs, through the superior cervical and the stellate ganglia. End-organs can show a reflex response after these afferent stimuli. Cerebral vessels, which are one type of “end-organ” for postganglionic sympathetic nerves, can also give a reflex response to sympathetic nerve stimulation.
The role of the superior cervical ganglion (SCG) and SG in cerebrovascular regulation has been confirmed by studies of reduced rCBF perfusion after SCG or SG stimulation [96-101] and increased perfusion after SCG or SG blockage. [102-104]
Our findings of a strong correlation between the number of painful spinal dysfunctions in the cervicothoracic-costal area in patients with neck pain and cerebral perfusion (although not independent of NDI scores in this sample) give some support to this somato-(spinal)-sympathetic theory. A model for this theory is presented in Figure 3.
Our study has important limitations. The sample size is somewhat small, particularly when creating 3 subgroups. This did not prevent the identification of strongly significant findings that did not appear to be spurious, in that they were consistent with our hypotheses. Nonetheless, replication of this study with a larger sample size would be warranted.
The analysis of SPECT results was only semiquantitative. However, the calculations performed on these SPECT data are strongly correlated to cerebral perfusion levels, and the rater analysis of color levels is based on expert levels of competence. More sophisticated quantitative analysis is strongly recommended for a replication of this work. The benefits of this would be to more strongly localize and statistically confirm the SPECT findings to ROIs and to identify these as specific cortical centers as opposed to the larger regions used here. In this regard, we would propose that our findings are preliminary and require confirmation with quantitative methods of rCBF measurement in future studies.
The identification of painful spinal dysfunction sites was conducted in an unblinded fashion and may have been susceptible to bias. It is important to note that this analysis was similar to a validated approach for the upper cervical spine by the assessor and was conducted blinded to NDI and SPECT findings, both of which would have seriously influenced the assessor had this order not been maintained. These findings could be obtained in a more blinded fashion and with 2 assessors in future studies.
We report here, for the first time, a correlation in patients with chronic neck pain between scores of self-rated disability (NDI), painful spinal joint dysfunction, and brain hypoperfusion. These findings further strengthen our understanding of the perceptions of patients with neck pain regarding their activities of daily living. They also provide preliminary evidence of a role for painful spinal joint dysfunction and a possible role for the sympathetic innervation of the cranial vasculature in the causal nexus of neck pain.