continuous
scanning of the migraine cycle over 30
days
and three
spontaneous attacks
Schulte LH, May A.
Abstract
Functional imaging using positron emission
tomography and later functional magnetic
resonance imaging revealed a particular
brainstem area that is believed to be
specifically activated in migraine during, but
not outside of the attack, and consequently has
been coined the 'migraine generator'. However,
the pathophysiological concept behind this term
is not undisputed and typical migraine
premonitory symptoms such as fatigue and
yawning, but also a typical association
of attacks to circadian and menstrual cycles,
all make the hypothalamus a possible regulating
region of migraine attacks. Neuroimaging studies
investigating native human migraine attacks
however are scarce and for methodological but
also clinical reasons there are currently no
studies investigating the last 24 h before
headache onset. Here we report a migraine
patient who had magnetic resonance imaging every
day for 30 days, always in the morning, to
cover, using functional imaging, a whole month
and three complete, untreated migraine attacks.
We found that hypothalamic activity as a
response to trigeminal nociceptive stimulation
is altered during the 24 h prior to pain onset,
i.e. increases towards the next migraine attack.
More importantly, the hypothalamus shows altered
functional coupling with the spinal trigeminal
nuclei and the region of the migraine generator,
i.e. the dorsal rostral pons during the preictal
day and the pain phase of native human migraine
attacks. These data suggest that although the
brainstem is highly linked to the migraine
biology, the real driver of attacks might be the
functional changes in hypothalamo-brainstem
connectivity.
Introduction
Among the more than 200 headache types,
migraine is the second most common headache
syndrome that affects between 12 and 14% of the
population. Migraine is predominantly a cycling
episodic disorder that manifests in attacks of
headache, photophobia, phonophobia and nausea
with a certain circadian rhythmicity. Extensive
research over the past 20 years has broadened
our understanding of the underlying mechanisms
and pathogenesis. Due to technological
improvements, brain imaging techniques have
gained importance in the quest to understand
cycling headache syndromes and speciÞc
emphasis has been placed to understand neuronal
activation in headache syndromes using
functional MRI.
Several independent functional studies have
established the crucial role of the brainstem in
acute and chronic migraine (Weiller et al.,
1995; Stankewitz et al., 2011) and the
hypothalamic area in tri- gemino-autonomic
headaches (May, 2005). A recent study reinforced
the speciÞc brainstem Þndings in
migraine by comparing brain responses during
trigeminal pain processing in migraine patients
with those of healthy control subjects
(Stankewitz et al., 2011). The main Þnding
was that the activity of the spinal trigeminal
nuclei in response to nociceptive stimulation
showed a cycling behaviour over the migraine
interval where the trigeminal activation level
increased signiÞcantly towards the next
migraine attack. However, this Þnding came
from a cohort study, where the time towards the
next attack was determined retrospectively, i.e.
per telephone contact after the experiment until
the next attack occurred. These data cannot
answer the question of whether the trigeminal
pain system is dysfunctional in itself or if
other structures modulate its activity.
From a clinical perspective, the
hypothalamus would be the most likely
'modulator' of the trigeminal pain system as the
many facets of a migraine attack, such as
yawning, fatigue and craving but also
circadian rhythmicity of at- tacks (Fox and
Davis, 1998; Fox, 2005; Alstadhaug et al., 2007;
Nascimento et al., 2014) could be best explained
when considering the biological role of the
limbic system. Using PET, one study reported
increased cerebral blood þow bilaterally
in the hypothalamus during an acute attack in
migraine patients (Denuelle et al., 2007) and
even more recently some hypothalamic activity
was reported very early in
nitroglycerin-triggered attacks (Maniyar et al.,
2014). Based on resting state data Moulton et
al. (2014) suggested that hypothalamic
connectivity with autonomic circuits and the
locus coeruleus in migraine may be altered. A
problem of all the above-mentioned studies is
the variance of cohort studies involving the
interpretation of averaged imaging data and the
fact that most studies investigate only a small
space of time of the migraine cycle, i.e. the
attack compared to a random day between attacks.
However, just as the migraine cycle spans
several days and contains up to Þve phases
(prodromes, aura, headache, resolution, and
recovery) (Blau, 1992), the trigeminal activ-
ity in migraine patients is not constant but
strongly variable (Stankewitz and May, 2007;
Stankewitz et al., 2011). Thus the ultimate
answer as to which areas of the brain and
brainstem might in fact generate migraine
attacks comes probably from individual data
which may give us a more differentiated view as
data differ from subject to subject depending on
the time to the next migraine attack.
Discussion
The main Þnding of this study is that
the hypothalamus, depending on the state of the
migraine cycle, exhibits an altered functional
coupling with the spinal trigeminal nuclei and
the region of the dorsal rostral pons. More
speciÞcally, the hypothalamus is
signiÞcantly more active within the last
24h preceding the onset of migraine pain and
shows the greatest functional coupling with the
spinal trigeminal nuclei, whereas during the
ictal state, the hypothalamus is functionally
coupled with the dorsal rostral pons, an area
that was previously coined 'the migraine
generator' (Bahra et al., 2001; Denuelle et al.,
2007).
In recent years migraine has primarily been
understood and discussed as a cyclic disorder
with physiological func- tioning and changing
activity of certain areas of the brain and
brainstem during different stages of the
migraine cycle (Weiller et al., 1995; Judit et
al., 2000; Stankewitz and May, 2007; Stankewitz
et al., 2011; Maniyar et al., 2014). Increased
hypothalamic activity in the hours preced- ing
migraine pain onset has only been shown once
imme- diately before NO-triggered migraine-like
headache (Maniyar et al., 2014). The authors
speculated that an activation in this area in
the preictal phase might represent a dysfunction
that could potentially modulate the top-down
inhibitory effect on the trigeminocervical
complex (Maniyar et al., 2014). It has to be
said that activation in the hypo- thalamic
region in this PET study was not seen, when all
preictal scans were compared against baseline
scans. Nevertheless, our data representing
spontaneous untreated attacks suggest
nitroglycerin-induced and spontaneous at- tacks
to be very similar, including imaging data.
Scanning the same person 30 days in a row
signiÞcantly reduces variance and allows
monitoring common and headache speciÞc
brain activations as a response to nociceptive
stimuli throughout the migraine cycle. One way
to analyse these data is to contrast different
states (e.g. interictal, preictal, ictal, and
postictal) with each other. Another opportunity
offered by this method is to use the migraine
states as a regressor and, over all 30 days,
compute the activity in the brain which
signiÞcantly follows this regressor. When
we contrasted the ictal with the interictal
state we found a strong and speciÞc
activation in the rostral pons, the region which
in earlier studies has been tightly linked to
attack generation (Weiller et al., 1995; Bahra
et al., 2001; Stankewitz et al., 2011). This
region was not prominently activated when
contrasting the preictal with the interictal
state, where the hypothalamus was
speciÞcally activated. In the postictal
phase, only the visual cortex showed stronger
activity as a response to pain compared to the
ictal phase. That the visual cortex activates as
a response to pain in interictal migraineurs is
well known (Boulloche et al., 2010). Taken
together with the Þndings from the
preictal and ictal phase as well as the
correlation contrast, this could point towards
increasing multisensory integration of visual
and nociceptive stimulation towards the next
migraine attack with a subsequent deactivation
during the pain phase of a migraine attack. This
phenomenon however could also represent a
baseline problem: in a situation of already
increased trigeminal nociceptive input the
pain-related visual cortex activity might
constitutively be higher leading to lesser
responses to additional painful stimulation. It
appears that each region is functionally highly
linked to a speciÞc phase in the migraine
cycle, a suggestion that is emphasized when
weighing all states over all cycles and all
days.
From a biological level it is probably more
important to understand whether such
phase-speciÞc activations have any
functional consequences and to explore possible
relationships between these areas. It is
therefore highly interesting that the
hypothalamic activity is not only phase locked
to the preictal phase but, in this phase, has a
strong functional coupling with the trigeminal
nuclei, which was not seen in the other phases.
We also found that the spinal trigeminal nuclei
increase activity towards the next migraine
attack and were thus able to replicate
Þndings from a previous study from our own
group (Stankewitz et al., 2011). The functional
coupling between the hypothalamus and the
trigeminal nuclei is signiÞcantly weaker
in the ictal phase, when the hypothalamus is
strongly coupled to the dorsal rostral pons.
This suggests that the activation in this area
in the preictal phase does not represent a mere
dysfunction but that the change in functional
connectivity of the hypothalamus with the tri-
geminal nuclei and the rostral pons drives the
different phases, probably by activating the
top-down inhibitory effect on the
trigeminocervical complex in the preictal phase
and activating the dorsal pons in the ictal
phase, eventually leading to the attack. The
current Þndings thus corroborate the
theory that the hypothalamus might be the true
generator of migraine attacks.
In conclusion, our data suggest the
hypothalamus to be the primary generator of
migraine attacks which, due to speciÞc
interactions with speciÞc areas in the
higher and lower brainstem, could alter the
activity levels of the key regions of migraine
pathophysiology.
