The key finding of this exploratory study is that patients with late-life MDD demonstrated significantly higher [18
F]FDDNP binding globally in the cortex with regional accentuation in the lateral temporal and PC regions when compared with healthy controls. Other regions, notably the anterior cingulate and mesial temporal, also showed higher [18
F]FDDNP binding when compared with controls, although the differences did not reach statistical significance. These findings, along with those of plasma studies of amyloid and [18
F]FDDNP binding correlates of anxiety and depression symptoms in patients diagnosed as having MCI and cognitively intact elderly individuals,8,9,34
indicate that neuronal injury, secondary to amyloid and tau, may represent a pathophysiologic pathway that, together with vascular compromise, may predispose elderly individuals to mood and related behavioral syndromes and disorders.
This pattern of high [18
F]FDDNP binding in the PC and lateral temporal regions is the same pattern described in a subgroup of 56 individuals without dementia.35
The pattern of binding differs from the pattern typically observed in patients diagnosed as having AD in whom [18
F]FDDNP binding is higher than controls throughout much of the neocortex and more pronounced in mesial temporal and partietal regions. In a more recent study,35
we compared [18
F]FDG–positron emission tomographic cerebral metabolic patterns in individuals without dementia in 3 subgroups defined according to their [18
F]FDDNP binding patterns. In that study, the [18
F]FDDNP subgroup with high lateral temporal and PC binding demonstrated heterogeneity in its [18
F]FDG–positron emission tomographic patterns, with a predominance of anterior frontal and anterior temporal hypometabolism, consistent with risk of mixed and/or other forms of dementia, including frontotemporal dementia.36
Our current results suggest that this high lateral temporal and PC binding in patients with MDD can be associated with different clinical patterns: some individuals show a clinical syndrome consistent with MCI but others demonstrate only depression symptoms. Longitudinal follow-up of these patients is necessary to evaluate the significance of these clusters and the clinical outcomes of these patient subgroups.37
The control and MDD groups in the current study showed a range of regional [18
F]FDDNP binding values (), suggesting that subgroups of the patients with MDD may show alternative patterns of regional binding, including the [18
F]FDDNP pattern HF/PA cluster. As shown in a separate study, this [18
F]FDDNP pattern is associated with an [18
F]FDG–positron emission tomographic pattern consistent with increased risk of AD (bilateral hypometabolism in the PA, temporal posterior cingulate, and dorsolateral prefrontal regions).36
Also, longitudinal data of patients with MCI indicate that this HF/PA pattern confers a high risk of cognitive decline after 2 years of follow-up.49
In these patients with MDD with the HF/PA pattern, depressive symptoms may be the initial manifestations of progressive neurodegenerative disease.
Neuroimaging studies, primarily MRI-based studies, have been used extensively to characterize the neuroanatomical and physiologic changes that underlie late-life MDD. Neuroanatomical approaches have revealed smaller brain volumes in key prefrontal, limbic, and subcortical regions in patients with MDD when compared with controls.2,3
Changes in gray matter density identified using sophisticated algorithms have shown increases and decreases when compared with controls.38,39
Magnetization transfer–based studies have revealed somewhat diffuse biophysical abnormalities in gray and white matter regions in the brains of patients with MDD.40,41
Magnetic resonance imaging–identified high-intensity lesions and abnormalities in fractional anisotropy detected using diffusion tensor imaging also are widespread in patients with MDD.50–52
Our current finding of relatively widespread increases in [18
F]FDDNP binding, with more marked involvement of some regions, is consistent with the findings of earlier neuroimaging studies that indicate that the biological underpinnings of late-life MDD are diffuse and involve multiple regions and neuronal circuits. Our MRI-facilitated positron emission tomographic image analysis of a subgroup also indicated widespread increase in [18
F]FDDNP binding, although the small sample size precluded regional measures of binding from becoming statistically significant.
The role of amyloid in the pathophysiologic manifestation of depression and dementia has received recent attention, although the findings sometimes are conflicting.8,11,53–58
Mayeux and coworkers54
reported that the risk of developing AD increased for individuals with higher plasma levels of Aβ42. Other reports suggest that lower Aβ42:Aβ40 ratios in the plasma identify individuals at risk for dementia, especially AD.53,55
Pomara and Murali Doraiswamy56
first proposed that increased platelet activation in patients with recurrent depression may lead to higher plasma levels of Aβ that, in turn, contribute to higher brain deposition of amyloid. Plasma studies by Sun and associates11
in patients diagnosed as having late-life MDD demonstrate that a subgroup of patients have lower levels of Aβ42 and higher Aβ40:Aβ42 ratios. In this study, patients with this plasma profile (high Aβ40:Aβ42 ratio) show impairments in memory and other cognitive domains comparable to the deficits observed in patients with AD. Pomara and Sidtis10
reported in their sample of patients with late-life MDD that patients with depression had higher levels of Aβ42 and Aβ42:Aβ40 ratios when compared with controls. Their data also suggest that higher Aβ42:Aβ40 ratios were associated with MRI-related brain abnormalities in patients with MDD. Despite this apparent discrepancy in the literature, both groups of investigators assert that changes in the Aβ42:Aβ40 ratios rather than changes in the absolute levels of either peptide are the relevant peripheral biological marker. These, together with other related observations, have led to the amyloidogenic theory of depression in late life that asserts that in a subgroup of patients with late-life MDD, perturbations of amyloid deposition and biology may be pathophysiologically relevant and may contribute to clinical and/or neuroimaging profiles suggestive of early dementia.56
Although the plasma results are intriguing, they are peripheral markers of neurobiology and reflect brain activity only indirectly.59–62
The precise relationship of plasma levels to brain neuronal levels in disease states and preclinical models has yet to be demonstrated.59–62
A study of PET images with validated imaging probes permits a more direct visualization of protein load and amyloid-induced injury in the brain.
An earlier study of PET images using [11
C]PIB and a small sample of patients with late-life MDD and controls identified higher [11
C]PIB brain retention in patients with MDD compared with controls. Higher [11
C]PIB retention was observed in several cortical areas comparable to the distribution seen in patients diagnosed as having AD.63
Of interest, higher [11
C]PIB retention was observed most noticeably in patients with MDD who concurrently met criteria for MCI (amnestic, nonamnestic, and mixed variety). Patients with MDD who did not meet the clinical criteria for MCI had [11
C]PIB brain retention parameters comparable to those for controls. In our sample, only 1 of the 20 patients with MDD met the criteria for MCI, and that patient’s [18
F]FDDNP binding parameters and distribution were similar to those of other patients with MDD. Our primary findings indicate that in patients with MDD who do not meet the established criteria for MCI, [18
F]FDDNP cortical binding is higher than in controls, indicating that brain neuropathologic aggregate deposition is present in MDD even in patients without discernible cognitive impairment.
Depression in late life is clinically and biologically heterogeneous. Current nosologic classifications follow empirical descriptive criteria and a somewhat arbitrary age cutoff for late-life disorders. Comparable to other behavioral and psychiatric classifications, late-life depression is diagnosed exclusively on clinical grounds with no acknowledgment of plausible etiologic considerations. Given this approach, it is not surprising that an entity or entities so defined will be biologically heterogeneous. Studies from several laboratories, including ours, have described multiple neuroimaging findings demonstrating several abnormalities in the brains of patients diagnosed as having late-life MDD. These findings include smaller brain volumes, biophysical abnormalities in multiple brain regions, and brain lesions of putative vascular origin. Although it would be premature to include late-life MDD in the category of amyloid and tauopathies, our current findings directly demonstrate that increased neuronal injury also can be correlated (or be secondary) to brain protein deposition, which may constitute another relevant biological mechanism in the underlying biology of MDD. At the cellular level, disparate mechanisms and pathways may converge and synergistically compromise neuronal structure and function even further.
Limitations of the current study include its preliminary nature, relatively small sample size of study groups, and the cross-sectional nature of the design. A longitudinal study with larger study samples is needed to establish the relationship of [18
F]FDDNP binding patterns, in vivo, to clinical outcomes. Also, we did not acquire genetic information regarding our patients and control groups. Although the prevalence of the APOE
allele, which is strongly associated with the risk of developing AD, is low in the general population, we are unable to comment on the association, if any, between APOE
status and [18
F]FDDNP binding in this sample. Also, although we have previously established that [18
F]FDDNP binding in AD is predominantly associated with tau aggregate deposition in the medial temporal lobe and largely reflects amyloid aggregates in other cortical areas, it is not possible to more precisely characterize the relative contributions of both of these proteins to [18
F]FDDNP binding in vivo.30
In conclusion, this is the first report, to our knowledge, to demonstrate increased [18F]FDDNP binding in focal brain regions suggesting higher amyloid and tau deposition in these brain areas in patients diagnosed as having late-life MDD. The pattern of binding, moreover, differs from that typically seen in patients with AD but it is consistent with that observed in controls at risk for dementia or patients with MCI. Neuronal injury secondary to higher protein deposition may represent a biologically plausible pathway to depression in late life. Longitudinal studies using large clinical samples are needed to determine whether higher [18F]FDDNP binding at baseline leads to AD over time. Also, [18F]FDDNP imaging via PET before and after treatment with conventional antidepressant therapy and antiamyloid agents will provide additional information regarding the change in protein neuropathologic deposition status after successful therapy with biologically relevant agents.