Biomarkers of alzheimer’s disease and mild cognitive impairment: are we there yet?

Abstract

Alzheimer’s disease (AD) may be caused by the age-dependent and progressive accumulation and deposition ofA -amyloid in brain—the “amyloid cascade hypothesis”(Hardy and Selkoe, 2002). A -peptides are cleaved fromthe transmembrane protein amyloid precursor protein (APP)by the proteases -secretase and -secretase. The secretedectodomain of APP released by -cleavage is APPs . Al-ternatively, -cleavage of APP within the A sequencethereby precludes A generation and results in secretion ofAPPs . There is considerable heterogeneity of secretedA -peptides, but A 42 is particularly amyloidogenic andfingered as an initial culprit in the pathologic cascadesleading to mild cognitive impairment (MCI) followed by theprogressive dementia diagnosed as AD. Proponents of theamyloid hypothesis are quick to point out that amyloid isnecessary but not sufficient for AD pathogenesis. Otherpathologies are equally important in neurodegeneration, in-cluding the formation of neurofibrillary tangles from hyper-phosphorylated tau. In fact, current neuropathologic criteriafor the diagnosis of high probability AD require abundantwidespread amyloid plaques and neurofibrillary tangles inbrain (National Institute on Aging and Reagan InstituteWorking Group on Diagnostic Criteria for the Neuropatho-logic Assessment of Alzheimer’s Disease, 1997) and incor-porate Braak and Braak (1991) and CERAD (Mirra et al.,1991) staging. Although multiple other pathologies areknown (neuronal and synaptic loss, gliosis, inflammation,oxidative injury, etc.) they are not considered in currentdiagnostic criteria and are placed downstream in the amy-loid cascade hypothesis.Operational clinical criteria define probable AD, or al-ternatively, possible AD if the dementia is atypical in onset,progression, or character or when an additional etiology ofdementia is considered (McKhann et al., 1984; AmericanPsychiatric Association, 1994; Knopman et al., 2001). “Def-inite” AD (more correctly high-probability AD) requirestissue diagnosis either from brain biopsy or necropsy—bothperformed infrequently in patients with dementia. The di-agnostic accuracy of probable and possible AD is 90 and 50%, respectively. In contrast, sensitivity, specificity, andpredictive values of non-AD dementias such as frontotem-poral dementia, Lewy body dementia, and vascular demen-tia are often far less than 90%. The identification of biomar-kers of AD and other dementias will add to ourunderstanding of their pathogenesis and to brain–behaviorrelationships and may be clinically useful in screening (toidentify high-risk subjects), to improve diagnostic accuracy(compared to necropsy), or to monitor efficacy of putativetherapies, particularly future potentially disease-modifyingtherapies now in preclinical phases of evaluation. Thebiomarker, or more likely a combination of markers, may bespecific to its function. For example in population screeninga marker with high sensitivity but low specificity may beapplied. However, for individual diagnostic accuracy highspecificity and positive predictive value are essential.Some putative biomarkers of AD, including some com-mercially available, are listed in Table 1. Various unvali-dated “at-home screening” instruments are not mentionedhere. Extensive critical reviews of biomarkers of AD maybe found elsewhere (Galasko, 2001; Knopman et al., 2001;Tuenissen et al., 2002). In general, markers may be dividedinto broad categories, including neuropsychometrics, struc-tural neuroimaging, functional neuroimaging, genomics,proteomics, or metabolomics. The latter terms refer to studyof the spectrum of genes, proteins, or metabolites in abiologic specimen, such as cells, tissues, or body fluids