Structural Brain Modifications in Primary Insomnia: Myth or Reality?

In comparison to other disorders of sleep, the number of brain imaging studies in chronic primary insomnia remains limited.^1,2 The earliest reports in this area utilized functional modalities. In a study using single photon emission computed tomography (SPECT) with ^99mTc-hexamethylenepropyleneamine oxime (^99mTc -HMPAO), patients displayed decreased perfusion in basal ganglia and various cortical areas during NREM sleep when compared to the same sleep stage in healthy controls.³ Nofzinger et al. then used positron emission tomography (PET) with ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) to investigate brain glucose metabolism in primary insomnia patients, during both NREM sleep and wakefulness.⁴ The main finding of their study was the demonstration that relative to control subjects, patients with primary insomnia showed less reduction of glucose metabolism during the transition from wakefulness to NREM sleep in the brainstem, hypothalamus, thalamus, limbic areas, and medial prefrontal cortex. This result is commonly referred to as supporting the hyperarousal hypothesis in insomnia.⁵ Using functional magnetic resonance imaging (fMRI), Altena et al.⁶ assessed brain responses during a verbal fluency task and found that chronic primary insomniacs showed less activation in medial and inferior prefrontal cortices than controls.

More recent neuroimaging studies of primary insomnia were exclusively dedicated to structural brain changes associated with this condition. In a pilot MRI study, Riemann et al. investigated volumetric differences in several brain regions of interest between chronic primary insomnia patients and control subjects.⁷ Only the hippocampus was reduced in volume in their small sample of 8 patients—a finding that did not survive correction for multiple comparisons. In order to reevaluate this hippocampal alteration in insomnia, two other MRI studies used a manual tracing of hippocampal volumes in 20 chronic primary insomnia patients, and both reported no significant decrease compared to controls.^8,9 Data from one of these two studies were reanalyzed to focus on another region of interest: the anterior cingulate cortex. Winkelmann et al.¹⁰ observed a volume increase of this area in patients, which was confirmed by the same authors on a second independent sample of 21 primary insomniacs, and interpreted as a compensatory response to chronic sleep disruption. Another study performed MRI scans in 24 chronic primary insomniacs and analyzed differences in gray and white matter across the whole brain using voxel-based-morphometry (VBM).¹¹ A significant reduction of gray matter volume was found in the left orbitofrontal cortex (with correction for multiple comparisons), which was interpreted in relationship with impaired cognitive functions, such as decision-making and problem-solving. A very recent study conducted by Spiegelhalder et al.¹² also utilized VBM on a sample of 28 primary insomnia patients. In contrast to the previous study,¹¹ image registration was performed by Spiegelhalder et al. using a toolbox called DARTEL (Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra), which improved the sensitivity of the analyses. Despite this methodological optimization, no differences between insomnia patients and controls were found in gray or white matter volumes.

In this issue of SLEEP, Joo and colleagues¹³ described the results of a VBM study using DARTEL registration in a sample of 27 patients with chronic primary insomnia, carefully screened by polysomnography (PSG) for the absence of other sleep disorders such as obstructive sleep apnea. In comparison to a group of 27 matched controls, patients showed decreases of gray matter concentration in the dorsolateral prefrontal cortex, primary sensorimotor cortex, and superior temporal gyrus. However, none of these changes remained significant after correction for multiple comparisons.

In addition to assessing group differences, the studies summarized above also attempted to find correlations between structural brain changes and clinical parameters, such as insomnia severity or duration. Some studies did not observe correlations between brain morphometry and insomnia clinical parameters.⁷ Other studies reported negative correlations between subjective insomnia severity and left orbitofrontal¹¹ and dorsolateral¹³ pre-frontal gray matter concentration or volume. Negative correlation with objective severity (e.g., PSG or actigraphy-defined wake after sleep onset) was observed with hippocampal volume,⁸ and left dorsolateral prefrontal and right primary sensorimotor gray matter concentrations.¹³ Positive correlations with objective and subjective insomnia severity were observed with anterior cingulate volume.¹⁰ No correlation was found with insomnia duration in three studies,^8,11,13 but two showed a negative correlation with hippocampal volume.^9,12 In the latter, lower hippocampal volumes were also associated with decreased performances in verbal and nonverbal memory tasks, thus linking structural brain changes to neuropsychological disturbances.⁹

A striking observation across these structural neuroimaging studies is the large variability of results. One possible explanation for such inconsistency resides in methodological differences in data analyses, involving techniques such as manually derived volumetric measurements^7–9 or various VBM modalities.^11–13 Another likely source of data variability stems from differences in the studied populations. Age, for instance, varied extensively across studies, from younger (e.g., 25-55 years)⁸ to older age ranges (e.g., 50-76 years).¹¹ PSG evaluation revealed that patients in some studies did not present a significant perturbation of objective sleep efficiency and architecture, thus leaning towards a paradoxical insomnia subtype,⁹ while other samples of patients showed a significant disruption of objective sleep latency and maintenance compared to control groups.^12,13 Correlations between structural brain changes and PSG or actigraphic measurements of clinical severity^8,10,13 illustrate the importance of taking into account the presence or absence of objectified alterations of sleep quality when conducting neuro-imaging research in insomnia.

Thus, the presence of neurostructural modifications in primary insomnia still remains an open question. The relative inconsistency of available studies precludes any definitive answer at the moment. It appears clear that such structural brain changes are strongly affected by data processing methods and importantly by the characteristics of the population sample. In this regard, age, disease severity, and PSG-defined sleep disruption are important factors to consider. The clinical heterogeneity of the studied samples might also contribute to the lack of robustness in some of the reported findings that did not survive correction for multiple comparisons. Future research should further investigate this issue by selecting large samples of carefully screened patients with clearly delineated clinical phenotypes. In addition, this field could also benefit from the application of other structural neuroimaging modalities, such as cortical thickness measurements and diffusion tensor imaging.

DISCLOSURE STATEMENT

Dr. Dang-Vu has indicated no financial conflicts of interest.

CITATION

Dang-Vu TT. Structural brain modifications in primary insomnia: myth or reality? SLEEP 2013;36(7):965-966.