Imaging of onset and progression of Alzheimer's disease with voxel-compression mapping of serial magnetic resonance images

Summary

Background

Early diagnosis and monitoring of the progression of Alzheimer's disease is important for the development of therapeutic strategies. To detect the earliest structural brain changes, individuals need to be studied before symptom onset. We used an imaging technique known as voxel-compression mapping to localise progressive atrophy in patients with preclinical Alzheimer's disease

Methods

Four symptom-free individuals from families with early-onset Alzheimer's disease with known autosomal dominant mutations underwent serial magnetic resonance imaging (MRI) over 5–8 years. All four became symptomatic during follow-up. 20 individuals with a clinical diagnosis of probable Alzheimer's disease and 20 control participants also underwent serial MR imaging. A non-linear fluid matching algorithm was applied to register repeat scans onto baseline imaging. Jacobian determinants were used to create the voxel-compression maps.

Findings

Progressive atrophy was revealed in presymptomatic individuals, with posterior cingulate and neocortical temporoparietal cortical losses, and medial temporal-lobe atrophy. In patients with known Alzheimer's disease, atrophy was widespread apart from in the primary motor and sensory cortices and cerebellum, reflecting the clinical phenomenology.

Interpretation

Voxel-compression maps confirmed early involvement of the medial temporal lobes, but also showed posterior cingulate and temporoparietal cortical losses at presymptomatic stage. This technique could be applied diagnostically and used to monitor the effects of therapeutic intervention.

Introduction

Alzheimer's disease is characterised histopathologically by senile amyloid plaques, neurofibrillary tangles, and neuronal loss. In established Alzheimer's disease, the cytoskeletal changes are widely distributed, and the medial temporal lobes and neocortical association areas are severely affected. Identification of how the disease starts and progresses is important for the development of therapeutic strategies aimed at prevention.¹ Necropsy studies have suggested that the entorhinal cortex and hippocampus are the sites of the earliest histological changes of Alzheimer's disease.² This suggestion accords with the prominent early memory deficits seen in Alzheimer's disease. However, cross-sectional necropsy studies can neither establish whether individuals with early pathological changes who were symptom-free before death would have gone on to develop Alzheimer's disease, nor how long these histological changes had been present before death. Moreover, when patients present with early memory impairment, evidence of more widespread cognitive changes are also apparent.3, 4

To resolve the issue of how the disease starts and progresses requires longitudinal follow-up of individuals from a presymptomatic stage by use of in-vivo techniques. To achieve this, previous studies have included patients with mild cognitive impairment. However, although these individuals do not fulfil the criteria for Alzheimer's disease, they already have symptoms of the disorder. An alternative group of individuals that could be studied longitudinally with a reasonable chance of identifying the earliest manifestations of the disease comprises the offspring of patients with autosomal dominant familial Alzheimer's disease. These individuals have a 50% risk of the disease and a relatively predictable age at onset. Comorbidity such as vascular disease is rare in these individuals because of their young age at disease onset. Furthermore, because of their known genetic risk of Alzheimer's disease, the diagnosis can be made with relative confidence while the patients are still alive and without histological confirmation. In longitudinal studies of individuals at risk of familial Alzheimer's disease, researchers have been able to identify in detail the emergence of cognitive deficits.3, 5 Although there are differences between autosomal dominant familial Alzheimer's disease and sporadic Alzheimer's disease, most notably the earlier age at onset in the familial form, the substantial clinical, pathological, and neuropsychological similarities support the use of familial Alzheimer's disease as a model of the disease that is amenable to presymptomatic study.

Non-invasive methods are required to track the physical progression of Alzheimer's disease, but the cellular hallmarks of the disease are currently beyond the resolution of in-vivo structural imaging techniques.6, 7 However, the macroscopic consequence of the cellular changes—ie, tissue loss—can be visualised with magnetic resonance imaging (MRI). But neuronal damage at the cellular level is accompanied by macroscopic structural readjustments: deformations, positional shifts, and volume loss, and this complex and dynamic process is inadequately captured with a single imaging snapshot of the brain. Serial imaging adds a fourth dimension—change over time—to three-dimensional MRI.

Sensitive methods are necessary to visualise and quantify subtle brain changes that might have occurred during the interval between scans. Digital subtraction of the serial imaging provides this sensitivity as long as accurate positional matching—registration—is first achieved. Rigid-body registration methods are now capable of bringing serial imaging into subvoxel alignment, allowing meaningful subtraction of an individual's later scan from their earlier scan. Registration that treats the brain as a rigid body can match the brain as a whole,⁸ and can be used to quantify whole-brain atrophy. However, because of structural readjustment, the sites of tissue loss cannot be accurately determined. A non-linear registration algorithm is needed⁹ to track local cerebral losses and deformations. A compressible viscous fluid model provides a plausible model of the structural changes associated with cerebral degeneration and has previously been used to visualise developmental change.¹⁰ We have shown that non-linear registration methods, incorporating this model, can also be used to match serial MRI scans from individuals with degenerative dementia.¹¹ This matching provides voxel-by-voxel maps of volume change, termed voxel-compression maps. We aimed to use this technique to localise regions of progressive atrophy in patients with presymptomatic Alzheimer's disease.