Cerebrospinal Fluid Aβ₄₂, Tau, and F₂-Isoprostane Concentrations in Patients With Alzheimer Disease, Other Dementias, and in Age-Matched Controls Open Access

Recent efforts have concentrated on determining markers to screen for increased susceptibility to Alzheimer disease (AD), to discriminate AD from other forms of dementia, and to quantify the efficacy of experimental therapeutics. Three major categories of markers have been investigated: genetic, neuroimaging, and biochemical. At a recent meeting convened by the National Institutes of Health, cerebrospinal fluid (CSF) Aβ₄₂ concentration, CSF tau level, and quantification of free radical–mediated damage to brain were discussed as potential biochemical markers for AD.1 

Others have quantified levels of CSF Aβ₄₂ and tau, and have addressed their utility, alone or in combination, as biomarkers of AD.2 F₂-Isoprostanes (F₂-IsoPs) have been employed widely as quantitative in vivo biomarkers of oxidative damage.3 F₂-Isoprostanes are exclusive products of free radical–mediated damage to arachidonic acid that are formed esterified to lipid and then are hydrolyzed and released into extracellular fluid. F₂-Isoprostanes are elevated in diseased regions of brain from definite AD patients compared to controls.4 Cerebrospinal fluid levels of F₂-IsoPs are elevated in probable AD patients compared to age-matched controls.5,6 In contrast, we have found that plasma and urinary F₂-IsoPs are not increased in probable AD patients.7 In combination, these data indicate that CSF F₂-IsoPs are biomarkers of oxidative damage to brain in AD. Since arachidonic acid is distributed among neuronal and nonneuronal elements in brain, F₂-IsoPs are not specific for neuronal oxidative damage. However, the value of measuring CSF F₂-IsoPs as biomarkers of AD relative to Aβ₄₂ and tau is not known. Here we have determined CSF Aβ₄₂, tau, and F₂-IsoP levels in patients with probable AD, other forms of dementia, and in age-matched control individuals.

All patients and control subjects were under care at Oregon Health Sciences University (Portland, Ore) or Vanderbilt University Medical Center (Nashville, Tenn). Probable AD patients were diagnosed according to National Institute of Neurologic and Communicative Disorders and Stroke-Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA) guidelines for probable AD.8 Diagnosis of dementia with Lewy bodies followed published criteria.9 Most patients were evaluated by neuroimaging studies, including all patients with normal pressure hydrocephalus. Diagnoses in patients with non-AD forms of dementia were based on best clinical judgment. Nondemented controls were 50 years of age or older, without significant cognitive or neurologic symptoms, and had a normal neurologic examination and Mini-Mental State Examination score of 28 or greater. Clinical Dementia Ratings,10 Mini-Mental State Examination scores,11 and Modified Ischemia Scores12 were determined according to established protocols.

Cerebrospinal fluid was obtained from the lumbar cistern of 37 individuals diagnosed as nondemented controls (n = 10), probable AD (n = 19), or other forms of dementia (n = 8). Diagnoses in demented patients without probable AD were dementia with Lewy bodies (n = 1), normal pressure hydrocephalus (n = 3), primary progressive aphasia (n = 3), and hippocampal sclerosis (n = 1). Results from clinical tests, age, sex, and disease duration are summarized in the Table.

Following informed consent, CSF was obtained from the lumbar cistern at both institutions as previously described, separated into aliquots, and stored at −80°C.5 All CSF samples had protein, glucose, and cell counts within the normal ranges. Aβ₄₂ and tau levels were determined by Athena Diagnostics, Worcester, Mass. F₂-Isoprostane levels were determined according to published methods using gas chromatography/negative ion chemical ionization mass spectrometry and [2H₄]-8-iso-PGF_2α as internal standard.5 

Statistical analysis was performed as described in the text using GraphPad Prism (San Diego, Calif). Nonparametric analysis of variance (ANOVA; Kruskal-Wallis test) was used to compare among the 3 groups of controls, AD, and non-AD dementia because the data sets did not have normal distributions. The Dunn multiple comparison test was used for post hoc comparisons to determine if significant differences existed between paired groups. In all tests α was set at .05.

Cerebrospinal fluid concentrations of Aβ₄₂, tau, and F₂-IsoPs were analyzed for all individuals (Figure 1). Similar to results from previous studies, CSF Aβ₄₂ levels decreased and CSF tau levels increased in probable AD patients compared to normal controls.13–15 Nonparametric ANOVA (Kruskal-Wallis test) was statistically significant for CSF Aβ₄₂ (P < .01) and tau (P < .05) levels when all 3 groups were compared. All 3 groups also were significantly different with respect to CSF F₂-IsoP levels (Kruskal-Wallis test, P < .01). Repeated pairs analysis was performed to determine significant differences between controls and probable AD patients, controls and patients with other dementias, or probable AD patients and patients with other dementias. Controls and probable AD patients differed significantly in their CSF concentrations of Aβ₄₂ (P < .01) and F₂-IsoPs (P < .01), but not tau (P > .05). Controls and patients with other dementias were not significantly different with respect to CSF Aβ₄₂, F₂-IsoP, or tau levels. Alzheimer disease patients and patients with other dementias had significantly different CSF levels of tau (P < .05), but not CSF Aβ₄₂ or F₂-IsoP levels. Cerebrospinal fluid levels of Aβ₄₂, tau, and F₂-IsoPs did not correlate with each other when comparing all individuals or just AD patients and controls. Moreover, the CSF concentrations of none of these 3 classes of molecules correlated with age, duration of illness, Clinical Dementia Rating, Mini-Mental State Examination score, Modified Ischemia Score, or number of ε4 alleles of APOE.

Six individuals, 3 AD patients, 2 other dementia patients, and 1 control, were taking vitamin E supplements in doses ranging from 400 to 2000 IU/day. A partially overlapping group of 12 individuals was taking vitamin C or multivitamin supplements, giving a total of 4 AD patients, 4 patients with other dementias, and 4 controls who were taking antioxidant vitamin supplements. Cerebrospinal fluid Aβ₄₂, tau, and F₂-IsoP levels for these 12 individuals were not significantly different from those of the other members of their respective groups.

The optimum value of CSF F₂-IsoP for discriminating AD from non-AD was greater than 25 pg/mL, as estimated by maximizing the sum of sensitivity (90%) and specificity (61%) for AD (n = 37). Combining CSF F₂-IsoP levels greater than 25 and the published discriminant value of Aβ₄₂ less than 1125 pg/mL,14 sensitivity and specificity for AD were 90% and 83% (n = 37), respectively (Figure 2). Finally, CSF Aβ₄₂, tau, and F₂-IsoP levels were combined into a 2-step algorithm for diagnosing AD. If CSF F₂-IsoP levels were less than 25 pg/mL, then the individual was classified as non-AD. If CSF F₂-IsoP levels were greater than 25 pg/mL, then CSF Aβ₄₂ and tau levels were considered according to the method of Galasko et al.14 Sensitivity and specificity values for AD versus non-AD with this algorithm were 84% and 89% (n = 37), respectively. For comparison, application of the CSF Aβ₄₂ and tau classification system of Galasko et al14 yielded 95% sensitivity and 50% specificity for AD in these 37 individuals.

This study tested the hypothesis that quantification of CSF F₂-IsoPs, when added to the classification system based on CSF Aβ₄₂ and tau levels, improved the laboratory diagnostic accuracy for AD. In our group of 37 probable AD patients, patients with other dementias, and controls, the classification system of Galasko et al,14 based on CSF Aβ₄₂ and tau levels alone, had high sensitivity but unacceptably low specificity for AD. Combined analysis of CSF Aβ₄₂ and F₂-IsoP levels, or F₂-IsoP levels followed by the classification system of Galasko et al,14 largely preserved sensitivity and improved specificity relative to classification with Aβ₄₂ and tau alone. An advantage of our analysis is that we included patients with other forms of dementia, and not just AD patients and controls, in the sensitivity and specificity calculations. However, our study is limited by a relatively small number of individuals in each group. Thus, the sensitivity and specificity values determined should be used to compare the different classification systems, and not as optimized values for classification. Future larger studies are needed to precisely define the sensitivity and specificity of simultaneous measurements of CSF Aβ₄₂, tau, and F₂-IsoPs in dementing illnesses.

Biomarkers that screen for increased susceptibility to AD, that assist in the differential diagnosis of dementia, or that may be used to assess response to therapy are important goals. Our study suggests that quantification of F₂-IsoPs in CSF may be a useful addition to existing biomarkers for probable AD patients. It is likely that the approach of using multiple biomarkers that reflect different aspects of AD pathogenesis, altered Aβ metabolism, disruption of the neuronal cytoskeleton, and increased free radical damage, among others, will provide the most accurate laboratory discrimination among controls and patients with different forms of dementia.

This work was supported by National Institutes of Health (Bethesda, Md) grants AG00774, AG16835, DK48831, GM15431, DK26657, and CA77839, as well as grants from the Alzheimer's Association, Chicago, Ill (Dr T. J. Montine), a Burroughs-Wellcome Clinical Scientist Award in Translational Research, Research Triangle Park, NC (Dr Morrow), and a Department of Veterans Affairs Career Development Award, Portland, Ore (Dr Quinn).