Shining a spotlight on headaches

Many headaches are worsened by light exposure. A new study uses a combination of human and rat experiments to suggest that a previously unknown retinal input to the thalamus may be important in such photophobia.

Now that the holiday season is over, many of us will be getting over memories of the painful mornings after the night before. On such mornings, the bright light of the day seems to be the worst thing for a throbbing headache. For migraine sufferers, the problem is even more acute and such photophobia is a critical aspect of their condition. But why should light exacerbate a nociceptive response and what is the neurological mechanism of this interaction?

A study in this issue¹ throws some metaphorical light on these questions. Using a combination of human behavioral and animal electrophysiology techniques, Noseda et al.¹ describe a previously unknown link between the visual and somatosensory system, showing that retinal afferents modulate the activity of neurons receiving nociceptive information from the dura mater (one of the meningeal coverings of the brain). These results suggest that a specific subset of thalamic neurons, which normally integrate information from the meninges and (presumably) cranial vasculature, can be modulated by visual inputs directly from retinal ganglion cells (RGCs). This data from rats is supported by observations that light aggravation of headaches in blind migraine sufferers depends on whether they can detect any light at all; in those individuals in which all vision had been lost, light does not exacerbate migraines. In contrast, blind patients who could detect light well enough to support apparently normal circadian entraining had migraine-associated photophobia. The authors propose that light can exacerbate headache in individuals with a partially intact light-responsive visual pathway.

The brain itself does not contain receptors that respond to noxious stimuli that result in painful sensations, including those from a headache. Instead, it is the outermost covering of the brain, the dura mater, and its associated blood vessels that are densely innervated by nociceptive neurons from the trigeminal and vagus cranial nerves^2,3. This trigeminal innervation occurs via small-diameter sensory neurons, which send information to the spinal trigeminal nucleus in the brainstem. Higher-order neurons relay this information to the thalamus, specifically targeting the ventrobasal complex, the lateral posterior and posterior nuclei, and to limbic associated thalamic nuclei⁴ (Fig. 1). The processing of nociceptive information in higher-order cortex is less clear; although parts of the somatosensory cortex, insula, anterior cingulate and prefrontal cortices are all commonly activated by painful stimuli, the manner in which this matrix of pain-associated areas are interlinked and differentially responsible for both the sensation and modulation of pain remains unresolved⁵. Lesions of primary somatosensory cortex and imaging of this region have produced conflicting results and it remains controversial whether the pathway via the ventro-posterior thalamic nuclei to somatosensory cortex is primarily involved in localization of pain or its more affective aspects^4,6. The more devastating aspects of the intense pain associated with conditions such as migraine are probably associated with limbic cortical areas of the insula and anterior cingulate cortex^7,8.

Figure 1: The proposed mechanism of light exacerbation of headaches through the convergence of the retinal input and nociceptive trigeminal afferents on the same thalamic projection neurons in the posterior group of the thalamus (Po).

The retinal afferents originate from RGCs or from ipRGCs and modulate the activity of neurons receiving nociceptive information through the spinal trigeminal nucleus from the meningeal covering of the brain (the dura mater). Noseda et al.¹ provide convincing anatomical and physiological evidence for these interactions. Blind migraine sufferers (where only ipRGCs, but no RGC are present) still experienced light aggravation during headaches. In those patients in which all vision had been lost, light did not exacerbate migraines.

Moving to the visual pathway, light is detected by retinal rod and cone photoreceptors and passed via the retinal interneuronal circuitry to RGCs, which in turn project to the central visual relays. Over a dozen subclasses of RGC have been identified⁹, and the most recent and intriguing of those are the RGCs that are independent of photoreceptors. These RGCs express the light-sensitive photopigment melanopsin and are referred to as intrinsically photosensitive RGCs (ipRGCs)¹⁰. These ipRGCs have been shown to be responsible for entraining the suprachiasmatic nucleus¹¹, an important node for determining circadian rhythms. In addition, a subpopulation of such ipRGCs innervate the pretectal nucleus and can mediate pupillary constriction responses¹².

Noseda et al.¹ found a third functional linkage in a nonvisual pathway that is mediated by these ipRGCs. The authors use elegant and convincing anatomical tract tracing methods to show that ipRGCs innervate the thalamic region that normally processes nociceptive input. Back filling of the retina from the thalamic nuclei, combined with immunohistochemistry, revealed the identity of these ipRGCs that project to the ventro-posterior thalamus. Electrophysiological recording of dural-driven nociceptive cells in animal experiments clearly showed that light was able to modulate the firing patterns of these cells, providing the functional linkage between the visual and sensory inputs. Although a more detailed analysis of the response characteristics and stimulus parameters in this interaction is required, the primary finding of an interaction was clear. To further establish their findings, the authors used juxtacellular dye labeling of physiologically characterized nociceptive/visually modulated thalamic cells, combined with anterograde retinal afferent labeling, to show retinal afferent innervation of these thalamic cells. Although only four thalamic cells were labeled in this way, the intimate relationship between the retinal terminals and thalamic neurons was clear. The juxtacellular-labeling technique also allowed thalamic projections to the cortex to be traced. These projections were not particularly well localized and the rather widespread terminations in multiple cortical regions limit any real understanding of how higher-order processing of this light-modulated nociceptive information might occur. As there was no quantification of these findings, it is hard to relate the demonstrably retinorecipient nociceptive thalamic neurons to specific cortical projection patterns.

The final part of this study attempts to link thalamic drive to cortical processing, and although the results are tantalizing, they are still incomplete. It is also not at all clear how thalamic input from a relatively small number of cells found in rats truly applies to humans, although the authors suggest that this input is the basis of the intense pain response associated with photophobia in migraine¹³. However, in the absence of a link between thalamic and cortical pain processing regions, it is still unclear what role these thalamic neurons might have in terms of cortical projections and pain perception. The rather diffuse thalamic projections to multiple cortical regions do not help in understanding how such information is processed at higher levels or in resolving the potential substrates for pain perception. It may be that the assumption that rats, as do humans, suffer from migraine is an over-ambitious one¹⁴.

These results therefore still leave many questions unanswered and provide much fodder for future work. For example, it would be interesting to tune the illumination to the physiological response properties of the ipRGCs^10,15 rather than being delivered at very low and very high intensities (500 lx and 50,000 lx). Such stimulation could provide even better evidence for the nature of the modulatory role of retinal afferents onto thalamic neurons by showing that thalamic response properties correlate well with properties of the ipRGCs. In the same vein, the future discovery of wavelength specific effects on partially sighted migraine sufferers might provide a relatively easy therapeutic approach to treat photo-exaggerated migraine. Would these patients show any form of selective wavelength-specific effects of light in terms of the light associated exacerbation of their migraine? These results suggest that an imaging study looking at the effects of differential illumination of migraine sufferers would be, quite literally, illuminating.