The clinical diagnosis of Alzheimer’s disease (AD) at the earliest clinical stages can be difficult. Given the variability in clinical threshold at which there is functional compromise in one’s social and professional life — a necessary component to AD diagnosis — there are several in-organic factors that affect early detection. Health professionals and biomedical scientists are searching for biomarkers, both biochemical or neuroimaging in nature, in order to objectively diagnose AD at various stages and provide invaluable prognostic information to the clinician. Whereas proteomic specialists have identified promising serum and CSF markers involved in AD pathophysiology (e.g. b-secretase, a key enzyme for amlyoid precursor protein cleavage), neuroradiologists have employed magnetic resonance in characterizing structural changes in the brain in vivo.

Hippocampal atrophy is implicated in the earliest stages of AD and even in preclinical stages, that is, in mild cognitive impairment (MCI). In fact, manual volumetric analysis is the “gold standard” to determine hippocampus volume and it is the best structural biomarker for AD. However, manual processing has its downsides. The procedure is a laborious and time-consuming task (>1 hr) significantly restricting its use as a routine diagnostic test. Moreover, it requires sophisticated knowledge of hippocampal and adjacent neuroanatomy and is subjective to operator error and inter-operator variability. In effort to solve these problems, Chupin and colleagues reported an automated method for hippocampus volumetry using segmentation. Briefly, the operator defines a bounding box over the amygdalohippocamal complex on a T1-weighted MR image and marks the center of both the hippocampus and amygdala. Then, the automated system identifies numerous anatomic landmarks (e.g. alveus and hippocampal sulcus) to segment the hippocampus and amygdala. Geometric algorithms are employed to calculate segmental atrophy.

Recently, Colliot and colleagues utilized the aforementioned automated method to discriminate between AD (25 patients), MCI (24 patients), and normal aging (25 elderly healthy controls). They performed three sets of analysis: group, individual, and subgroup. The group analysis revealed statistically significant reductions in both AD and MCI patients. The individual analysis first established a training set (75% of each group) to cluster individuals into one of the three groups based on volumetric data. The data was tested with the remainder 25% of each group. When compared to elderly healthy controls, 84% of AD patients and 73% for MCI patients were correctly classified. 69% of AD patients were correctly classified compared to MCI patients. Lastly, the investigators matched for age and sex and obtained approximately similar correct classification rates.

Colliot’s study marks a novel approach using an automated method to hippocampal volumetry to discriminate AD, MCI, and normal aging and is one of a few reports to provide accuracy figures (i.e. correct classification rates). However, there are several limitations in extrapolating this study’s findings to the general population and in routine diagnostic testing.

As I reported in 2006, ideally a diagnostic marker in general meets eight conditions:

• 1. detects a fundamental feature of the disease with high sensitivity and specificity;
• 2. validated in post-mortem confirmed cases;
• 3. standardized with sound bioinformatics;
• 4. specific for the disease compared with related disorders;
• 5. reliable in many testing environments/labs;
• 6. noninvasive;
• 7. simple to perform; and
• 8. inexpensive.

Automated volumetric analysis for AD and MCI meets most of the above criteria; however, there is not enough data to comment on its reliability (and certainly not expense given the proprietary nature of the algorithm and analysis software). This study was a small scale prospective evaluation with a total of 25 AD patients. Moreover, a single operator performed all initializations and therefore data on inter-operator variability is unknown.

The patient recruitment/selection methods are not clearly delineated; the article reads “the patients were selected from a database of patients prospectively recruited” at their medical center. Factors such as advertising method may influence the patient population’s degree of illness. For instance, if newspaper, TV, or internet ads recruited individuals with AD or MCI, they may have less impairment (and resultant hippocampus changes) than those recruited from nursing homes.

Concerning the MCI group, the authors fail to identify if other cognitive domains were also impaired (i.e. language, visuospatial skills, or executive function). Individuals with amnestic MCI-multiple domain subtype will progress to AD and death more rapidly, and may express greater hippocampal changes than single-domain patients.

The future may lead us to an era of objective testing for AD including gene arrays, protein chips, and structural MRIs with automated processing. The clinical implications of AD/MCI biomarkers are far and wide. Although there are currently no FDA approved drugs for MCI and/or AD prevention, there are several trials underway using cholinesterase inhibitors, COX-2 inhibitors, and vitamin E. Neuroimaging can be employed as end-points in phase II/III clinical trials of disease-modifying therapies. With additional studies, models may be constructed that may predict disease progression from MCI to AD and across various stages of AD.

Imaging in neurodegenerative disorders is still in its infancy. Further research is needed with greater power in each group, detailed patient clinical/lab sub-classifiers (i.e. single vs. multiple domain MCI, ApoE4 carrier status, drug therapy), long-term intervals, and tests for inter-operator variability.

References

CHUPIN, M., MUKUNABANTUMBAKULU, A., HASBOUN, D., BARDINET, E., BAILLET, S., KINKINGNEHUN, S., LEMIEUX, L., DUBOIS, B., & GARNERO, L. (2007). Anatomically constrained region deformation for the automated segmentation of the hippocampus and the amygdala: Method and validation on controls and patients with Alzheimer’s disease NeuroImage, 34 (3), 996-1019 DOI: 10.1016/j.neuroimage.2006.10.035

Colliot, O., Chetelat, G., Chupin, M., Desgranges, B., Magnin, B., Benali, H., Dubois, B., Garnero, L., Eustache, F., & Lehericy, S. (2008). Discrimination between Alzheimer Disease, Mild Cognitive Impairment, and Normal Aging by Using Automated Segmentation of the Hippocampus Radiology, 248 (1), 194-201 DOI: 10.1148/radiol.2481070876

Gauthier, S., Reisberg, B., Zaudig, M., Petersen, R., Ritchie, K., Broich, K., Belleville, S., Brodaty, H., Bennett, D., & Chertkow, H. (2006). Mild cognitive impairment The Lancet, 367 (9518), 1262-1270 DOI: 10.1016/S0140-6736(06)68542-5

Hunderfund, A., Roberts, R., Slusser, T., Leibson, C., Geda, Y., Ivnik, R., Tangalos, E., & Petersen, R. (2006). Mortality in amnestic mild cognitive impairment: A prospective community study Neurology, 67 (10), 1764-1768 DOI: 10.1212/01.wnl.0000244430.39969.5f

Lakhan, S. (2006). Schizophrenia proteomics: biomarkers on the path to laboratory medicine? Diagnostic Pathology, 1 (1) DOI: 10.1186/1746-1596-1-11

Petersen RC, & Negash S (2008). Mild cognitive impairment: an overview CNS Spectr, 13 (1), 45-53