Dr Ben Seymour, Computational and Biological Learning Lab, Trumpington Street, Cambridge CB2 1PZ


Center for Information and Neural Networks, National Institute for Information and Communications Technology (NICT), 1-3 Yamadaoka, Suita, Osaka 565-0871, Japan.

bjs49 AT / seymour AT

F1000 reviews (paywall)



Pain highlights: August.

The relative contribution of peripheral versus central mechanisms to chronic neuropathic pain following peripheral nerve injury has been a long-standing debate. A couple of important papers were published this month, which illustrate the different roles for played by various peripheral nerve subtypes.

Laiche Djouhri, from Sally Lawson’s lab in Bristol studied rats with partial nerve ligations. It’s commonly thought that one of the main mechanisms of chronic pain in nerve lesions arises from central (dorsal horn and higher) plasticity following loss of input from the Wallerian degeneration of axotomised peripheral neurons. Djourhi et al show that the nerves that survive might be just as important. They show that C-, Ad-, and Ab-nociceptors all display abnormal activity including spontaneous abnormal firing, abnormal thresholds and altered membrane dynamics. This suite of abnormal signaling in intact neurons seems likely to contribute to the diverse sensations that patients report following nerve injury. 

However, what is still not clear to what extent this activity might be responsible for the maintenance of chronic pain, beyond the precipitating injury. Ayano Nakao from Fusao Kato’s lab in Jikei University show that TRPV1 expressing c-fibre’s may have a specific role in the central pathogensis of neuropathic pain. They show that ablating c-fibres with post-natal capsaicin has little effect on mechanical allodynia induced by spinal nerve ligation. However, they observe significant reduction in the potentiation of the connectivity between lateral parabrachial nucleus and capsular portion of the central nucleus of the amygdala (a major downstream pathway for lamina 1 dorsal horn neurons that receive c-fibre inputs from the periphery). This shows that this pathway seems to be specifically dependent on c-fibre integrity. It also highlights the need to understand exactly what specific aspects of pain behaviour is mediated by the parabrachial-amygdala pathway -  a question that could be addressed by neuroimagers.  

Finally, on a more clinical note, evidence for what may be a new chronic pain disorder was published in Neurology this month. Christopher Klein and colleagues at the Mayo clinic have discovered that pain is commonly associated with voltage-gated potassium channel (VGKC) autoimmunity. VGKC autoimmunity has been one of the most interesting new chapters in neurology in the last decade – with a spectrum of clinical disorders spanning peripheral motor nerve hyperexcitability, neuromyotonia and most recently, limbic encephalitis and psychosis. Klein et al reviewed the cases of positive VGKC assays who had been neurologically assessed, and found a large number of patients with pain as an isolated phenomenon. Typical features were neuropathic and nociceptive (and often misdiagnosed as psychogenic pain or fibromyalgia), and pain was often accompanied with associated features such as abnormal sweating and thermal sensation – consistent with peripheral nociceptor hyperexcitability. Sera was subtyped positive for LGI1-IgG and CASPR2-IgG. Importantly, pain often resolved with immunotherapy. This defines a new and treatable clinical pain syndrome, and as word gets out and pain neurologists start sending of the antibodies, may turn out to be more common than previously thought (as was the case with VGKC encephalitis).  


Laiche Djouhri, Xin Fang, Stella Koutsikou, Sally N. Lawson. Partial nerve injury induces electrophysiological changes in conducting (uninjured) nociceptive and nonnociceptive DRG neurons: Possible relationships to aspects of peripheral neuropathic pain and paresthesias. Pain. 153 (2012) 1824–1836. 

Ayano Nakao, Yukari Takahashi, Masashi Nagase, Ryo Ikeda, Fusao Kato. Role of capsaicin-sensitive C-fiber afferents in neuropathic pain-induced synaptic potentiation in the nociceptive amygdala. Molecular Pain 2012, 8:51

Christopher J. Klein, Vanda A. Lennon, Paula A. Aston, Andrew McKeon, Sean J. Pittock. Chronic pain as a manifestation of potassium channel-complex autoimmunity. Neurology. Published online before print August 15, 2012


Is chronic pain a neural connectivity disorder?

Interesting things are happening in the field of chronic pain. Since fMRI emerged  20 years ago, attention has progressed from probing the specific roles of individual brain regions towards the study of the roles of networks of brain areas. Broad connectivity-based approaches emerged from the discovery of discrete functional networks, characterized by strong intra-regional correlated low frequency oscillatory activity, and corresponding anti-correlation with other brain networks: best known is the default mode network (DMN). Many neuroscientists have been rather skeptical about the DMN, since its functional definition (in behavioural terms) is inherently and necessarily vague - it is said to support ‘intrinsic, self-referential information processing’. But the DMN has caught the attention of pain neuroscientists, who are acutely aware that chronic pain is often most apparent at rest, and related phenomenologically to whatever intrinsic, self-referential processing might be, occurring without any obvious ascending nociceptive sensory input. 

Recently, a number of studies have shown that resting-state DMN activity is abnormal in chronic pain patients (or various aetiologies), and the degree of abnormal connectivity has been shown to correlate with pain duration, severity, spontaneous fluctuations, natural resolution and treatment. Last year, Baliki and colleagues showed that this activity relates specifically to abnormally high frequency oscillations (0.12 - 0.20Hz), as measured by fMRI blood-oxygen-level-dependent (BOLD) signal, in the medial prefrontal cortex (mPFC). In the latest installment of what has been a fascinating program of research from this group, they now show that abnormal connectivity between mPFC and nucleus accumbens (NAcc) predicts the future chronification of back pain in patients presenting with a new but short history of pain. Critically, this suggests that abnormal connectivity plays a causal role in chronic pain, and is not merely a downstream marker of chronic pain. This gives what seems to be the clearest neuroscientific insight yet into the pathogenesis of chronic pain in these patients to date. It’s also interesting to speculate on the behavioural significance of this connectivity: these areas are thought to support experienced-based integration and control of reward and pain, so hints at a possible disorder of adaptive motivational control.

Abbendum: We recently discussed the issue of temporal predictability and causality here in our journal club. Whereas the study suggests causality (ie. is consistent with), it doesn't prove it. This is because it might be the case that an as yet unidentified factor both causes back pain chronification and independently, abnormal connectivity. The latter could then be mechanistically (causally) unrelated, albeit correlated, with chronic back pain. 

The cortical rhythms of chronic back pain. Baliki MN, Baria AT, Apkarian AV J Neurosci. 2011 Sep 28; 31(39): 13981-90 PMID: 21957259 DOI: 10.1523/JNEUROSCI.1984-11.2011. This is a modified from one of my evaluations.



Von Economo neurons - will they turn out to have a role in pain?

Von Economo neurons enjoy a level of privilege and exclusivity afforded to no other cell. This is because they inhabit only the brightest brains in the animal kingdom – higher primates, whales, dolphins, elephants to name a few. This has rendered them experimentally largely untouchable, and allowed them to run free in the imagination of neuroscientists, playing center stage in many grand theories of mind and consciousness. 

But for how much longer? Thanks to some careful and sophisticated microscopy, Evrard, Forro and Logethetis have now found them in the macaque brain [1], opening the doorway to future functional studies. Although much less abundant than in humans and the great apes, there seems little doubt that they are indeed there, most abundantly in the anterior insula. 

So what makes them so special? Apart from being present only in the very smart and very social, it’s the fact that the anterior insula and anterior cingulate cortex are two regions strongly implicated in things like social behaviour, empathy, self awareness, and even suicide. Furthermore, they are very big projection neurons, so clearly have the capacity to send ultra-fast information across broad areas of cortex. But this is all circumstantial, because until now no-one has been able to record from or stimulate them in awake behaving animals. 

So what’s this got to do with pain? Possibly not much, but of course the anterior insula and anterior cingulate are also the brain regions most reliably activated in studies of pain. And pain is sometimes argued to be the ‘highest’ form of consciousness, and related emotions such as empathy are crucial to social bonding. So could they have some specific role in pain – perhaps sending a global message to other parts of the brain when pain is detected? 

In a thoughtful commentary [2], Seth and Critchley suggest that they might have a role in interoception – the more general sense of the integrity of the body spanning thermal, autonomic, and pain sensation. In particular, they suggest it might mediate a pattern of recurrent connectivity with subcortical regions (such as the periaqueductal grey), as part of a hierarchical model of interoceptive perception. Such a model mimics more general theories of cortical function and perception, popular in domains such as vision, and something we amongst others have suggested might also be applicable to pain [3]. At first glance a perceptual role for Von Economo neurons might seem unlikely, otherwise why would we not see big neurons doing something similar in the visual system? But then again, interoception and pain are not much like vision, because they have significant intrinsic salience (usually aversive). Indeed, it is it’s capacity for things like pain to dominate our consciousness and command multiple autonomic and cognitive resources that might just require a system that rapidly evaluates and responds to stimuli in a special way. Unfortunately, however, it’s difficult to study pain in the macaque, but it will still be fascinating to see what happens during recordings from other stimuli and tasks. 

[1] Evrard et al. Von Economo Neurons in the Anterior Insula of the Macaque Monkey 
Neuron, Volume 74, Issue 3, 482-489, 10 May 2012 

[2] Seth and Critchley. Will Studies of Macaque Insula Reveal the Neural Mechanisms of Self-Awareness? Neuron, Volume 74, Issue 3, 423-426, 10 May 2012 

[3] Ben Seymour and Ray Dolan. Emotion, Motivation, and Pain. In Wall and Melzack's Textbook of Pain. 2012. Ch. 17. Eds. Kotzenburg and McMahaon. Elsevier. In press.


How much of the pain BOLD response is actually pain?

In a recent article, Moulton and colleagues illustrate that interpreting blood oxygen level-dependent (BOLD) responses in studies of human pain is a bit more complicated than often recognised. In particular, their study probes two important factors:

1) painful and non-painful thermal sensation have a non-identical cortical representation, and this lack of clear disambiguation in previous studies may have led to over interpretation of the role of S2 in pain, 

2) the time course of BOLD response is not necessarily a clear-cut haemodynamic impulse response – there is emerging evidence of early and late phase responses, which may subserve different functions. Of note, perceptual ratings correlate better with late phase response.

This latter point is especially interesting, since it really isn't obvious what at an information processing level the pain system would be doing during a late phase that isn't/can't be done early. This should stimulate some interesting new research, as well as appeal to other methodologies such as electroencephalography (EEG) and magnetoencephalography (MEG) which give better temporal precision.

[1] Moulton EA, Pendse G, Becerra LR, Borsook D. BOLD responses in somatosensory cortices better reflect heat sensation than pain. J Neurosci. 2012 Apr 25; 32(17): 6024-31


The missing cortex?

In a new paper {1}, Mazzola and colleagues present a further account of their well-known studies of electrical stimulation of the cortex (see ref {2}), in which they identify sensations of pain arising from a region of cortex around the medial parietal operculum and posterior insula cortex. 

Their research raises a fundamental question -– is this really the missing pain cortex? Do these rather small, inconsistent, and sometimes unpleasant sensory as opposed to frankly painful experiences reflect the activity of the elusive 'SIII'. It's a fascinating question and forces us to ask ourselves what it would take for us to conclude that this area is actually the long lost cortex. One the one hand, it seems likely that many nociceptive-specific ascending pathways project here (see ref {3} for a review) and stimulation here seems sufficient to cause to pain, but on the other, it's hardly the robust and convincing result you might expect for a sensory and emotional experience that can overwhelm our consciousness. 

So what should we expect to characterize a pain cortex? Is it specificity for pain? Or the necessity and sufficiency to support conscious appreciation of pain? Or the engagement in complex computations involved in processing pain? Or all three of these? The region is certainly not exclusive for pain, but there may be some degree of specificity (see ref {4} for a recent investigation). Necessity is hinted at by lesion studies (for an example, see ref {5}), and sufficiency is implied by this study. Perhaps what is most lacking is a clear demonstration that the area performs a specific computational function involved in the perception of pain. And herein lies a possible clue to the difficulty finding it, since pain is a comparatively simple, primitive sensation (compared to vision or hearing, for instance), there may not be an awful lot of complex computation to do, hence necessitating only a small bit of (shared) cortex. The only obvious alternative account is that pain perception has a genuinely distributed coding across two or more regions of cortex. If this is the case, perhaps simultaneous stimulation of these regions (for example, anterior cingulate and posterior insula cortex) might produce more pain than either alone.


1. Mazzola L, Isnard J, Peyron R, Mauguière F. Stimulation of the human cortex and the experience of pain: Wilder Penfield's observations revisited. Brain. 2012 Feb; 135(Pt 2): 631-4. PMID: 22036962 DOI: 10.1093/brain/awr265

2.Somatotopic organization of pain responses to direct electrical stimulation of the human insular cortex. Mazzola L, Isnard J, Peyron R, Guénot M, Mauguière FPain 2009 Nov; 146(1-2): 99-104 PMID: 19665303 DOI: 10.1016/j.pain.2009.07.014

3.Interoception: the sense of the physiological condition of the body. Craig ADCurrent opinion in neurobiology 2003 Aug; 13(4): 500-5  PMID: 12965300
4.The human operculo-insular cortex is pain-preferentially but not pain-exclusively activated by trigeminal and olfactory stimuli. Lötsch J, Walter C, Felden L, Nöth U, Deichmann R, Oertel BGPloS one 2012; 7(4): e34798  PMID: 22496865 DOI: 10.1371/journal.pone.0034798
5.Isolated insular infarction eliminates contralateral cold, cold pain, and pinprick perception. Birklein F, Rolke R, Müller-Forell WNeurology 2005 Nov 8; 65(9): 1381 PMID: 16275823 DOI: 10.1212/01.wnl.0000181351.82772.b3