ORIGINAL CONTRIBUTION. Correlation of Longitudinal Cerebrospinal Fluid Biomarkers With Cognitive Decline in Healthy Older Adults

Transcription

1 ORIGINAL CONTRIBUTION Correlation of Longitudinal Cerebrospinal Fluid Biomarkers With Cognitive Decline in Healthy Older Adults Erik Stomrud, MD; Oskar Hansson, MD, PhD; Henrik Zetterberg, MD, PhD; Kaj Blennow, MD, PhD; Lennart Minthon, MD, PhD; Elisabet Londos, MD, PhD Background: Abnormal cerebrospinal fluid (CSF) biomarker levels predict development of Alzheimer disease with good accuracy and are thought to precede cognitive deterioration. Objective: To investigate whether changes in CSF biomarker levels over time in healthy older adults are associated with a concurrent decline in cognitive performance. Design: Retrospective analysis of longitudinal CSF biomarker levels and clinical data. Setting: A combined academic dementia disorder research center and dementia clinic. Participants: Thirty-seven cognitively healthy older volunteers (mean age, 73 years). Main Outcome Measures: Longitudinal CSF total tau protein, hyperphosphorylated tau protein 181, and -amyloid 1-42 protein levels and cognitive assessments at baseline and at follow-up 4 years later. Results: Low levels of CSF -amyloid 1-42 protein at follow-up were associated with decreased delayed word recall score on the Alzheimer Disease Assessment Scale cognitive subscale (r s = 0.437, P.01) and with slower results on A Quick Test of Cognitive Speed (r s = 0.540, P.001). Individuals with a decrease during the 4-year study of 15% or more in CSF -amyloid 1-42 protein level performed worse on the Alzheimer Disease Assessment Scale cognitive subscale delayed word recall (z= 2.18, P.05) and A Quick Test of Cognitive Speed (z= 2.35, P.05) at follow-up. An increase over time of 20% or more in CSF hyperphosphorylated tau protein 181 level correlated with slower results on A Quick Test of Cognitive Speed at follow-up (z= 2.13, P.05). Furthermore, the presence of the APOE-ε4 (OMIM ) allele was associated with a greater longitudinal decrease in CSF -amyloid 1-42 protein level ( 2 =10.47, P.05) and with a higher CSF total tau protein level at follow-up ( 2 =8.83, P.05). No correlation existed between baseline CSF biomarker levels and baseline or follow-up cognitive scores. Conclusions: In this group of healthy older adults, changes in CSF biomarker levels previously associated with Alzheimer disease correlated with a decline in cognitive functions. Changes in CSF biomarker levels may identify early neurodegenerative processes of Alzheimer disease. Arch Neurol. 2010;67(2): Author Affiliations: Clinical Memory Research Unit, Department of Clinical Sciences, Malmö University Hospital, Lund University, Malmö (Drs Stomrud, Hansson, Minthon, and Londos), Kalmar County Hospital, Kalmar County Council, Kalmar (Dr Stomrud), and Institute of Neuroscience and Physiology, Department of Neurochemistry and Psychiatry, Sahlgrenska University Hospital, Gothenburg (Drs Zetterberg and Blennow), Sweden. ALZHEIMER DISEASE (AD) IS A neurodegenerative disorder with a disease onset that is believed to occur several decades before cognitive symptoms can be detected. 1 A long preclinical phase exists before cognitive functions begin to decline, leading to mild cognitive impairment (MCI) 2 and subsequently to clinical dementia. 3,4 Cognitive functions first affected in AD are thought to be episodic memory, perceptual speed, and executive function, 5 and this decline has been detectable up to 9 years before fulfillment of AD criteria. 6,7 Several diagnostic biologic markers, including cerebrospinal fluid (CSF) biomarkers, have effectively predicted conversion from MCI to AD dementia. 8,9 Hence, in the new revised research criteria for AD (National Institute of Neurological and Communicative Disorders and Stroke Alzheimer s Disease and Related Disorders Association), CSF biomarker levels are considered supportive features. 4 The predictive power of these diagnostic markers in a pre-mci stage, before any measurable cognitive impairment, has not yet been fully investigated. Nevertheless, a few studies have identified such markers in these early stages of cognitive decline. The morphologic changes most commonly associated with AD are extracellu- 217

2 lar senile plaques composed of -amyloid 1-42 protein (A 42) and intracellular neurofibrillary tangles composed of hyperphosphorylated tau protein 181 (P-tau 181). 3,15,16 Furthermore, levels of the CSF biomarkers total tau protein (T-tau), P-tau 181, and A 42 demonstrate the best ability so far to distinguish and predict AD, especially when they are used in combination. 8,15,16 The longitudinal stability of these CSF biomarker levels in individuals with AD or MCI seems, in part, to be dependent on the length of time between the CSF collections. In the short term, all 3 biomarkers tend to remain stable, 17,18 which also seems to be the case for CSF tau protein levels in the long term according to most published studies, with some exceptions. 26,29 In contrast, CSF A 42 levels predominantly remained stable in short-term studies 17,18,24,27,28 (duration 3 years) but decreased in longterm studies 21,22 (duration 3 years), except for one reported increase. 29 In addition, performance on cognitive tests has not been correlated with CSF biomarker levels in studies 20,22,23 investigating CSF biomarker stability. Most studies in which high biomarker stability was reported were performed among individuals with AD dementia or MCI, which would suggest that the actual transition of CSF biomarkers to pathologic levels in the development of AD occurs in preclinical stages. However, the pattern of changes in CSF biomarker levels over time in preclinical stages of AD is unknown. Similarly, studies investigating preclinical longitudinal changes in CSF biomarker levels in relation to cognitive functions are sparse. Therefore, we investigated whether changes occur in CSF biomarker levels over time in cognitively healthy older volunteers and whether a tendency to change is associated with cognitive functions. The study was approved by the regional ethics committee at Lund University, Lund, Sweden. The individuals gave their written consent to participate at each assessment occasion. METHODS SUBJECTS The study subjects were cognitively healthy older adults recruited by the memory clinic at Malmö University Hospital, Malmö, Sweden, to constitute a healthy control group in dementia studies. Recruitment occurred in 2002 through advertisements, and the individuals underwent a thorough examination at baseline (medical history, cognitive testing, and somatic and psychiatric examination [including computed tomography]). Inclusion criteria at baseline were intact activities of daily living functions, no memory complaints, and cognitive test results within the expected normal range. Exclusion criteria were active physical or mental disease that could affect the cognitive status, including advanced pathologic findings on computed tomography, fulfillment of criteria for AD 30 or other dementia types, and fulfillment of criteria for MCI. 2 Within 6 months from baseline, the included individuals underwent CSF collection and apolipoprotein E (APOE) genotyping. Two follow-up cognitive assessments were then performed after 3 and years, with the latter including a second CSF collection. The same basic cognitive test battery was used at the time points for CSF collection, including the Mini-Mental State Examination (MMSE) 31 and the Alzheimer Disease Assessment Scale cognitive subscale (ADAS-cog), 32 with the addition of A Quick Test of Cognitive Speed (AQT) 33 and clock drawing at the time point for the second CSF collection. Cognitive test results or fulfillment of criteria for dementia disorder or MCI at follow-up did not affect participation in this study. However, individuals were excluded at follow-up if they failed to complete any of the CSF collections or an entire cognitive test battery. The remaining individuals at follow-up were included in this longitudinal study and were available for statistical analyses. COGNITIVE TESTS The test battery consisted of the MMSE and the ADAS-cog, with the addition of AQT and clock drawing at follow-up. At the study clinic, these were the initial screening tests used for referrals at that time. Of the tests, the MMSE is known to be insensitive in early development of AD. 34 In comparison, the ADAS-cog is more sensitive, especially if delayed word recall is included. 32 Therefore, the 85-point version (ADAS-cog85) was used in this study. In AQT, 33 participants are asked to name 40 figures in 3 subsets according to their color, form, and color-form, respectively, while the time is monitored. It originates from the 1935 Stroop color-word test, and slower speed has been associated with AD. 33 The 10-Point Clock Test 35 was used for assessment of clock drawing. The total test scores and their individual task scores, representing cognitive functions changed in early AD, are used for calculations. CSF COLLECTION AND ASSAYS The same procedures were performed in the first and second CSF collections and analyses. Lumbar puncture was performed in a sitting position. The CSF samples were obtained in the L3/L4 or L4/L5 interspaces. After disposal of the first 1 ml of CSF, consecutive 10-mL samples were collected in plastic (polypropylene) tubes to avoid absorbance of A 42 by the tube wall. All CSF samples were mixed gently to avoid possible gradient effects. No CSF sample contained more than 500 erythrocytes/µl. The CSF samples were centrifuged at 2000g and 4 C for 10 minutes to eliminate cells and other insoluble material and were then immediately frozen and stored at 80 C pending biochemical analyses without being thawed or refrozen. At the same time, serum samples were collected. The CSF samples were analyzed for A 42, T-tau, and P-tau 181 levels using xmap technology (INNOBIA AlzBio3 kit; Innogenetics, Ghent, Belgium) and the same batch of reagents. 36 The results are given in nanograms per liter. A random sample of stored baseline CSF was analyzed together with the follow-up CSF to secure concordant assay values between the 2 analysis occasions. LONGITUDINAL CSF ANALYSES For longitudinal CSF analyses, the individual change in each CSF biomarker level from baseline to follow-up was calculated. However, these data could be subject to methodological problems, especially in individuals with stable CSF values around 0. Therefore, we aimed to identify those individuals with substantial changes in CSF biomarker levels at follow-up compared with baseline that were leading to CSF levels in a more pathologic direction. The decision about what constitutes a pathologic direction was determined by the changes in CSF biomarker levels seen in AD compared with controls (ie, increasing T-tau and P-tau 181 levels and decreasing A 42 level). 15 We also sought to use a relative change value for each individual instead of an absolute change value because of the wide range of CSF biomarker levels. We dichotomized the group for each CSF biomarker in relation to the relative change, with cut- 218

3 off levels at 10% change, 15% change, or 20% change in a pathologic direction. Dichotomization according to the same inverted percentage change for each CSF biomarker (ie, in the nonpathologic direction) was also performed to control for the fact that the correlations are not a result of measuring outliers in general. STATISTICAL ANALYSIS Statistical analysis was performed using commercially available software (SPSS, version for Windows; SPSS Inc, Chicago, Illinois). Nonparametric tests were used when the variables were not normally distributed. Spearman rank correlation coefficient was used to test the degree of correlation between CSF biomarker levels and cognitive test results, as well as the influence of age. Wilcoxon signed rank test was used to test longitudinal changes within the same variable (CSF biomarker levels, MMSE score, and ADAS-cog85 score). Mann- Whitney test was used when one of the variables was dichotomized (longitudinal CSF biomarker data, dropout, and sex). Kruskal-Wallis test was used to test the influence of the presence of the APOE-ε4 allele on quantitative measures. The level of significance was set at P.05. RESULTS DEMOGRAPHICS, COGNITIVE TEST, AND CSF RESULTS Thirty-seven of 54 individuals who had undergone successive CSF collections and cognitive assessments at baseline were included in the study. The demographics of the study population at baseline and at follow-up are given with the cognitive test results on a group level in Table 1. The mean CSF biomarker levels at baseline and at follow-up are given in Table 2. Dropout analysis shows that reasons for not participating in the follow-up were somatic disease (6 individuals), death (4 individuals), and other (3 individuals), whereas 4 individuals participated in the follow-up but failed to complete the second CSF collection. The mean baseline ADAS-cog85 delayed word recall score of 13 dropouts was 1.4 points higher than that of 41 participants (z= 2.34, P.05). No difference was seen for age, sex, APOE genotype, or CSF biomarker levels. There was no difference in cognition at follow-up between the groups who completed or did not complete the second CSF collection (P.05). At follow-up, none of the participants fulfilled criteria for MCI or dementia. CROSS-SECTIONAL CORRELATIONS OF CSF BIOMARKER LEVELS AND COGNITION At baseline, no correlations between CSF biomarker levels and any of the 7 cognitive test results were seen (Table 3). At follow-up, lower CSF A 42 levels significantly correlated with higher ADAS-cog85 delayed word recall scores (r s = 0.437, P.01) (Figure 1) and ADAScog85 total score (r s = 0.330, P.05) and with slower results on AQT subset color (r s = 0.385, P.05), AQT subset form (r s = 0.540, P.001) (Figure 2), and AQT subset color-form (r s = 0.422, P.01). If the delayed word Table 1. Demographics and Cognitive Test Results Variable (n=54) Follow-up Age, mean (SD), y 73 (8) 73 (8) 77 (8) Sex, No. Female Male APOE-ε4 heterozygote 16 (1) 9 (1) 9 (1) (homozygote), No. MMSE score Mean (SD) 29.2 (1.0) 29.3 (1.0) 28.3 (1.8) Range ADAS-cog85 score Mean (SD) 7.4 (3.6) 7.0 (3.6) 8.1 (4.6) Range AQT subset score, median (25th-75th percentiles) Color (22-29) Form (29-37) Color-form (56-78) 10-Point Clock Test score Mean (SD) (0.5) Range Time between CSF collections, mean (SD), mo (1.6) Abbreviations: ADAS-cog85, 85-point version of the Alzheimer Disease Assessment Scale cognitive subscale; APOE-ε4, apolipoprotein E-ε4; AQT, A Quick Test of Cognitive Speed; CSF, cerebrospinal fluid; ellipsis, not applicable; MMSE, Mini-Mental State Examination. recall score was excluded from the ADAS-cog85 total score, the correlation with CSF A 42 level no longer existed (P.05). Neither CSF T-tau level nor P-tau 181 level correlated with any of the cognitive test results at followup. No additional effect was seen if the ratio of A 42 level to P-tau 181 level or the ratio of A 42 level to T-tau level was used. Older age correlated with worse results on the MMSE (r s = 0.388, P.05), ADAS-cog85 (r s =0.435, P.01), and AQT (r s =0.480, P.01) at follow-up, whereas no age correlation was observed with CSF biomarker levels. Neither the MMSE nor clock drawing correlated with CSF biomarker levels. The individuals who were APOE-ε4 allele carriers had significantly higher T- tau levels at follow-up compared with noncarriers ( 2 =8.83, P.05). No other correlations were seen at baseline or at follow-up between the presence of the APOE-ε4 allele and CSF biomarker levels or cognitive test results. Sex did not correlate with CSF biomarkers levels or with cognitive test results. LONGITUDINAL CORRELATIONS OF CSF BIOMARKER LEVELS AND COGNITION No significant longitudinal changes in CSF biomarker levels or ADAS-cog85 delayed word recall scores were observed (P.05). However, MMSE scores were a mean of 1 point lower (z= 2.8, P.01) and ADAS-cog85 total scores a mean of 1 point higher (z= 2.18, P.05) at follow-up. CSF biomarker levels and longitudinal changes in CSF biomarker levels were not associated with cognitive test results at follow-up (P.05). However, sig- 219

4 Table 2. CSF Biomarker Levels Variable, ng/l (n=54) Follow-up T-tau level Mean (SD) (214.2) (227.3) (189.3) Median (25th-75th percentiles) ( ) ( ) ( ) P-tau 181 level Mean (SD) 65.3 (19.2) 63.4 (19.0) 63.0 (27.0) Median (25th-75th percentiles) 59.0 ( ) 56.9 ( ) 56.0 ( ) A 42 level Mean (SD) (200.1) (177.9) (162.3) Median (25th-75th percentiles) ( ) ( ) ( ) Abbreviations: A 42, -amyloid 1-42 protein; CSF, cerebrospinal fluid; P-tau 181, hyperphosphorylated tau protein 181; T-tau, total tau protein. Table 3. Correlation of CSF Biomarker Levels With Cognitive Test Results (Not Corrected for Multiple Comparisons) Spearman Rank Correlation Coefficient Follow-up Variable T-tau Level P-tau 181 Level A 42 Level T-tau Level P-tau 181 Level A 42 Level MMSE total score ADAS-cog85 Delayed word recall a Total score b AQT subset score c Color b Form d Color-form a 10-Point Clock Test score c Abbreviations: A 42, -amyloid 1-42 protein; ADAS-cog85, 85-point version of the Alzheimer Disease Assessment Scale cognitive subscale; AQT, A Quick Test of Cognitive Speed; CSF, cerebrospinal fluid; ellipsis, not applicable; MMSE, Mini-Mental State Examination; P-tau 181, hyperphosphorylated tau protein 181; T-tau, total tau protein. a P.01. b P.05. c Only performed at follow-up. d P.001. nificant correlations emerged when groups were dichotomized according to the occurrence of a substantial change in a pathologic direction among CSF biomarker levels. At follow-up, 11 individuals with a decrease of 15% or more in CSF A 42 level performed significantly worse on the ADAS-cog85 delayed word recall (z = 2.18, P.05) (Figure 3A) compared with the others. Seven individuals had decreases as high as 20% or more in CSF A 42 level, but these individuals were significantly slower on AQT subset color-form (z= 2.35, P.05) (Figure 3B). Six individuals with an increase of 20% or more in CSF P-tau 181 level were significantly slower on AQT subset color (z= 2.13, P.05) (Figure 3C). The individuals with pathologic changes both in CSF A 42 level and CSF P-tau 181 level (n=4) did not differ in cognition at follow-up from the others. No significant differences in cognitive performance were observed among individuals with a substantial change in the CSF biomarker levels in a nonpathologic direction compared with the others. Individuals with the APOE-ε4 allele had a greater longitudinal decrease in CSF A 42 level than those without the allele ( 2 =10.47, P.05). COMMENT This longitudinal study of cognitively healthy older adults suggests that low CSF A 42 levels are associated with worse episodic memory and cognitive speed. The crosssectional associations were not present 4 years earlier, and the CSF A 42 levels at this time were unassociated with the cognitive outcome at follow-up. Instead, individuals with an extensive pathologic change in A 42 levels and P-tau 181 levels over the 4 years between CSF collections had worse episodic memory and cognitive speed at follow-up. Hence, in these healthy older adults, CSF biomarker levels and changes over time previously associated with AD seem to correlate with cognitive functions known to decline in early development of AD. 3,15,16,21,22 The known early affected cognitive functions, episodic memory and cognitive speed, correlate with crosssectional and longitudinal deviance in CSF biomarker levels in this study. 5-7 In contrast, no correlations between CSF biomarker levels and MMSE scores or its episodic 220

5 Follow-up CSF Aβ42 Level, ng/l Follow-up CSF Aβ42 Level, ng/l Follow-up ADAS-cog85 Delayed Word Recall Score Figure 1. Follow-up delayed word recall scores on the 85-point version of the Alzheimer Disease Assessment Scale cognitive subscale (ADAS-cog85) in relation to follow-up cerebrospinal fluid -amyloid 1-42 protein (CSF A 42) levels, with the median values and the interquartile ranges shown (37 individuals). Circle defines outliers between 1.5 and 3 times the interquartile range Follow-up AQT Subset Form Score, s Figure 2. Distribution of follow-up scores on A Quick Test of Cognitive Speed subset form (AQT Subset Form) in relation to follow-up cerebrospinal fluid -amyloid 1-42 protein (CSF A 42) levels (37 individuals). The regression curve is indicated (r s = 0.540, P.001). Follow-up ADAS-cog85 Delayed Word Recall Score No Yes CSF Aβ42 Level Decrease 15% AQT Subset Color-Form Score, s No Yes CSF Aβ42 Level Decrease 20% AQT Subset Color Score, s No Yes CSF P-tau 181 Level Increase 20% Figure 3. Median values and interquartile ranges are shown, and circles define outliers between 1.5 and 3 times the interquartile range; asterisks define outliers greater than 3 times the interquartile range. A, Follow-up delayed word recall scores on the 85-point version of the Alzheimer Disease Assessment Scale cognitive subscale (ADAS-cog85) according to a decrease in cerebrospinal fluid -amyloid 1-42 protein (CSF A 42) level of 15% or more (11 yes and 26 no). B, Follow-up scores on A Quick Test of Cognitive Speed subset color-form (AQT Subset Color-Form) according to a decrease in CSF A 42 level of 20% or more (7 yes and 30 no). C, Follow-up scores on AQT subset color according to an increase in CSF hyperphosphorylated tau protein 181 (P-tau 181) level of 20% or more (6 yes and 31 no). memory task were seen. However, this is not surprising because the MMSE is known to lack sensitivity in early stages of AD 34 and has fewer memory items compared with the ADAS-cog85. In this study, some individuals demonstrated single outlying test results as summarized in Table 1. However, a thorough investigation of these individuals rejects any fulfillment of criteria for cognitive disorders, and the performance drop occurs in separate and random cognitive tasks only. Furthermore, age did not correlate with CSF biomarker levels at baseline or at follow-up and has not been adjusted for in the study. Inclusion of the worst baseline cognitive status in the dropout group might lead to a type II error; therefore, greater differences would be required in this study to be considered significant. Of the CSF biomarkers in this study, primarily CSF A 42 level correlates with cognitive test results. This correlation is supported by previous findings in cognitively healthy older adults in which primarily CSF A 42 level has been shown to correlate with future cognitive decline and development of dementia. 10,12,14 In addition, baseline CSF A 42 levels in the present study population predicted subjective memory impairment affecting quality of life at the 3-year follow-up. 11 In contrast to the subjective measurement at the 3-year follow-up, CSF A 42 level correlates with more objective measurements herein at the year follow-up. Therefore, it may be that the association of CSF A 42 level with cognitive functions at this possible pre-mci stage shifts and moves closer toward possible MCI and AD. The findings in the present study are further in alignment with the decrease in CSF A 42 level and the increase in CSF T-tau and P-tau 181 levels that have been reported by numerous CSF studies in MCI and patients with very mild AD and dementia. 8,9,15,16,37 However, in MCI studies, 8,9,15,16 all 3 CSF biomarker levels seem to be affected and associated with the risk and rate of progression to AD dementia. In addition, several previous studies 8,10,11,13,15 on CSF biomarker levels and cognition in healthy older adults or patients with MCI report improved results when combining CSF A 42 level and P-tau 181 level or T-tau level 221

6 in ratios or as cutoff levels, but this was not seen herein when ratios (A 42/P-tau 181 level and A 42/T-tau) were used. Instead, CSF A 42 and CSF tau protein might represent different mechanisms in the course of events in the development of AD; hence, the CSF biomarker levels will have different correlation patterns with other diagnostic markers for AD. For example, follow-up CSF T-tau and P-tau 181 levels in the present study population have been correlated with slowing of electroencephalogram results, whereas CSF A 42 level was not. 38 Furthermore, pathologic tau and CSF tau levels have been more closely related to cognitive performance in stages of AD dementia and progressive MCI compared with pathologic A 42 level. 25,39 Therefore, the present study findings suggest that the CSF A 42 level more concurrently parallels the cognitive decline in preclinical stages. Longitudinal changes at the group level were only seen in measurements (MMSE and ADAS-cog85 total scores) that did not significantly correlate with any other measurements. Therefore, the cross-sectional correlation at follow-up of the CSF A 42 level with episodic memory and cognitive speed suggests a concurrent decrease in CSF A 42 level and impairment in these cognitive functions on an individual basis. The mechanism behind the decrease in CSF A 42 level is not fully known, but aggregation into plaques, decreased A 42 production with neuron loss, and increased clearance from the CSF have been proposed. 40 Likewise, the possible underlying biologic mechanism behind the observed relationship between CSF A 42 level and cognition can only be speculated. However, a connection of this unknown mechanism to neuron loss would be reasonable because most neuropathologic evidence on early stages of AD implicates neuron loss as a major feature correlated with cognitive performance. 41 Longitudinal CSF studies 17,18,21,22,24,27,28 in AD dementia have reported inconclusive results for CSF A 42 in which stable or decreasing levels are predominant, as already described. Furthermore, the longitudinal stability of CSF biomarker levels within AD samples has not been shown to correlate with the extent of cognitive decline. 20,22,23,42 Therefore, the transition to pathologic CSF biomarker levels seems not to occur in the clinical stages of AD. Instead, the present study findings suggest a preclinical occurrence of the pathologic transition of CSF A 42 and P-tau 181 levels. In alignment with this preclinical theory, a decrease in CSF A 42 levels over time has been correlated with longitudinal loss of hippocampal volume in individuals with MCI 24 and has been shown to moderately predict conversion from MCI to AD dementia. 16 Furthermore, P-tau 231 level has been proposed to increase over time after adjustment for ventricular volume measured using magnetic resonance imaging. 26 For now, it is difficult to draw any specific conclusions from the single correlation between increased P-tau 181 level over time and slower cognitive speed. The correlation between the longitudinal change in CSF A 42 level and the presence of the APOE-ε4 allele has been reported in other studies 21,43-45 among control subjects and among individuals with AD and may further indicate that the longitudinal change in CSF A 42 levels in this study could be of a pathologic nature. In light of this, the absence of any relationship between the presence of the APOE-ε4 allele and follow-up CSF A 42 levels is hard to explain. Similarly, it is unknown why CSF T-tau level instead correlates with the presence of the APOE-ε4 allele at follow-up, as studies 19,43-46 among individuals with AD have reported inconclusive results for CSF T-tau level and APOE genotype, whereas a correlation with APOE genotype almost always exists for CSF A 42 level. Any explanations for the observed relationship between the CSF T-tau level and APOE-e4 allele in this study would be speculative and should be addressed by future studies. In conclusion, CSF biomarker levels associated with the development of AD in several studies 8,9,15,16 correlated with a decline in cognitive functions in this group of cognitively healthy older adults. Furthermore, more extensive CSF biomarker level changes in a pathologic direction over time correlate with a decline in the same cognitive functions. Changes in CSF biomarker levels may identify individuals experiencing pre-mci and possible early neurodegenerative processes of AD. Accepted for Publication: August 10, Correspondence: Erik Stomrud, MD, Clinical Memory Research Unit, Department of Clinical Sciences, Malmö University Hospital, Lund University, S Malmö, Sweden (erik.stomrud@med.lu.se). Author Contributions: All authors had full access to all the data in the study, and Dr Stomrud takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Stomrud, Blennow, Minthon, and Londos. Acquisition of data: Stomrud, Zetterberg, Blennow, and Minthon. Analysis and interpretation of data: Stomrud, Hansson, Zetterberg, Blennow, Minthon, and Londos. Drafting of the manuscript: Stomrud. Critical revision of the manuscript for important intellectual content: Hansson, Zetterberg, Blennow, Minthon, and Londos. Statistical analysis: Hansson. Obtained funding: Stomrud, Blennow, and Minthon. Administrative, technical, and material support: Zetterberg, Blennow, and Minthon. Study supervision: Hansson and Londos. Financial Disclosure: None reported. Funding/Support: This study was supported by unconditional grant from the Swedish Research Council (Dr Blennow), Region of Skåne (Dr Stomrud), County Council of Kalmar (Dr Stomrud), Foundation for Old Servants, and Alzheimer Foundation Sweden. Additional Contributions: Eva Falk Langebro, Tarja Tikkanen, and the rest of the clinical trial group at the Neuropsychiatric Clinic, Malmö University Hospital, provided administrative support and management of the cerebrospinal fluid samples. REFERENCES 1. Price JL, Morris JC. Tangles and plaques in nondemented aging and preclinical Alzheimer s disease. Ann Neurol. 1999;45(3): Petersen RC, Stevens JC, Ganguli M, Tangalos EG, Cummings JL, DeKosky ST; Report of the Quality Standards Subcommittee of the American Academy of Neurology. Practice parameter: early detection of dementia: mild cognitive impairment (an evidence-based review). Neurology. 2001;56(9): Blennow K, de Leon MJ, Zetterberg H. Alzheimer s disease. Lancet. 2006;368(9533):

7 4. Dubois B, Feldman HH, Jacova C, et al. Research criteria for the diagnosis of Alzheimer s disease: revising the NINCDS-ADRDA criteria. Lancet Neurol. 2007; 6(8): Bäckman L, Jones S, Berger AK, Laukka EJ, Small BJ. Cognitive impairment in preclinical Alzheimer s disease: a meta-analysis. Neuropsychology. 2005;19 (4): Amieva H, Jacqmin-Gadda H, Orgogozo JM, et al. The 9 year cognitive decline before dementia of the Alzheimer type: a prospective population-based study. Brain. 2005;128(pt 5): Nestor PJ, Scheltens P, Hodges JR. Advances in the early detection of Alzheimer s disease. Nat Med. 2004;10(suppl):S34-S Hansson O, Zetterberg H, Buchhave P, Londos E, Blennow K, Minthon L. Association between CSF biomarkers and incipient Alzheimer s disease in patients with mild cognitive impairment: a follow-up study. Lancet Neurol. 2006;5(3): Mattsson N, Zetterberg H, Hansson O, et al. CSF biomarkers and incipient Alzheimer disease in patients with mild cognitive impairment. JAMA. 2009;302 (4): Fagan AM, Roe CM, Xiong C, Mintun MA, Morris JC, Holtzman DM. Cerebrospinal fluid tau/ -amyloid 42 ratio as a prediction of cognitive decline in nondemented older adults. Arch Neurol. 2007;64(3): Stomrud E, Hansson O, Blennow K, Minthon L, Londos E. Cerebrospinal fluid biomarkers predict decline in subjective cognitive function over 3 years in healthy elderly. Dement Geriatr Cogn Disord. 2007;24(2): Skoog I, Davidsson P, Aevarsson O, Vanderstichele H, Vanmechelen E, Blennow K. Cerebrospinal fluid -amyloid 42 is reduced before the onset of sporadic dementia: a population-based study in 85-year-olds. Dement Geriatr Cogn Disord. 2003;15(3): Li G, Sokal I, Quinn JF, et al. CSF tau/a 42 ratio for increased risk of mild cognitive impairment: a follow-up study. Neurology. 2007;69(7): Gustafson DR, Skoog I, Rosengren L, Zetterberg H, Blennow K. Cerebrospinal fluid -amyloid 1-42 concentration may predict cognitive decline in older women. J Neurol Neurosurg Psychiatry. 2007;78(5): Blennow K. CSF biomarkers for mild cognitive impairment. J Intern Med. 2004; 256(3): de Leon MJ, Mosconi L, Blennow K, et al. Imaging and CSF studies in the preclinical diagnosis of Alzheimer s disease. Ann N Y Acad Sci. 2007;1097: Blennow K, Zetterberg H, Minthon L, et al. Longitudinal stability of CSF biomarkers in Alzheimer s disease. Neurosci Lett. 2007;419(1): Andreasen N, Minthon L, Vanmechelen E, et al. Cerebrospinal fluid tau and A 42 as predictors of development of Alzheimer s disease in patients with mild cognitive impairment. Neurosci Lett. 1999;273(1): Andreasen N, Minthon L, Clarberg A, et al. Sensitivity, specificity, and stability of CSF-tau in AD in a community-based patient sample. Neurology. 1999;53 (7): Sunderland T, Wolozin B, Galasko D, et al. Longitudinal stability of CSF tau levels in Alzheimer patients. Biol Psychiatry. 1999;46(6): Huey ED, Mirza N, Putnam KT, et al. Stability of CSF -amyloid 1-42 and tau levels by APOE genotype in Alzheimer patients. Dement Geriatr Cogn Disord. 2006; 22(1): Tapiola T, Pirttila T, Mikkonen M, et al. Three-year follow-up of cerebrospinal fluid tau, -amyloid 42 and 40 concentrations in Alzheimer s disease. Neurosci Lett. 2000;280(2): Andreasen N, Vanmechelen E, Van de Voorde A, et al. Cerebrospinal fluid tau protein as a biochemical marker for Alzheimer s disease: a community based follow up study. J Neurol Neurosurg Psychiatry. 1998;64(3): de Leon MJ, DeSanti S, Zinkowski R, et al. Longitudinal CSF and MRI biomarkers improve the diagnosis of mild cognitive impairment. Neurobiol Aging. 2006; 27(3): Brys M, Pirraglia E, Rich K, et al. Prediction and longitudinal study of CSF biomarkers in mild cognitive impairment. Neurobiol Aging. 2009;30(5): de Leon MJ, Segal S, Tarshish CY, et al. Longitudinal cerebrospinal fluid tau load increases in mild cognitive impairment. Neurosci Lett. 2002;333(3): Mollenhauer B, Bibl M, Trenkwalder C, et al. Follow-up investigations in cerebrospinal fluid of patients with dementia with Lewy bodies and Alzheimer s disease. J Neural Transm. 2005;112(7): Zetterberg H, Pedersen M, Lind K, et al. Intra-individual stability of CSF biomarkers for Alzheimer s disease over two years. J Alzheimers Dis. 2007;12(3): Bouwman FH, van der Flier WM, Schoonenboom NS, et al. Longitudinal changes of CSF biomarkers in memory clinic patients. Neurology. 2007;69(10): McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer s Disease. Neurology. 1984;34(7): Folstein MF, Folstein SE, McHugh PR. Mini-Mental State : a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975; 12(3): Rosen WG, Mohs RC, Davis KL. A new rating scale for Alzheimer s disease. Am J Psychiatry. 1984;141(11): Jacobson JM, Nielsen NP, Minthon L, Warkentin S, Wiig EH. Multiple rapid automatic naming measures of cognition: normal performance and effects of aging. Percept Mot Skills. 2004;98(3, pt 1): Tombaugh TN. Test-retest reliable coefficients and 5-year change scores for the MMSE and 3MS. Arch Clin Neuropsychol. 2005;20(4): Manos PJ, Wu R. The ten point clock test: a quick screen and grading method for cognitive impairment in medical and surgical patients. Int J Psychiatry Med. 1994;24(3): Olsson A, Vanderstichele H, Andreasen N, et al. Simultaneous measurement of -amyloid 1-42, total tau, and phosphorylated tau (Thr181) in cerebrospinal fluid by the xmap technology. Clin Chem. 2005;51(2): Snider BJ, Fagan AM, Roe C, et al. Cerebrospinal fluid biomarkers and rate of cognitive decline in very mild dementia of the Alzheimer type. Arch Neurol. 2009; 66(5): Stomrud E, Hansson O, Minthon L, Blennow K, Rosen I, Londos E. Slowing of EEG correlates with CSF biomarkers and reduced cognitive speed in elderly with normal cognition over 4 years [published online ahead of print May 5, 2008]. Neurobiol Aging doi: /j.neurobiolaging Goedert M, Spillantini MG. A century of Alzheimer s disease. Science. 2006;314 (5800): Andreasen N, Blennow K. CSF biomarkers for mild cognitive impairment and early Alzheimer s disease. Clin Neurol Neurosurg. 2005;107(3): Price JL, Ko AI, Wade MJ, Tsou SK, McKeel DW, Morris JC. Neuron number in the entorhinal cortex and CA1 in preclinical Alzheimer disease. Arch Neurol. 2001; 58(9): Josephs KA, Whitwell JL, Ahmed Z, et al. -Amyloid burden is not associated with rates of brain atrophy. Ann Neurol. 2008;63(2): Sunderland T, Mirza N, Putnam KT, et al. Cerebrospinal fluid -amyloid 1-42 and tau in control subjects at risk for Alzheimer s disease: the effect of APOE ε4 allele. Biol Psychiatry. 2004;56(9): Prince JA, Zetterberg H, Andreasen N, Marcusson J, Blennow K. APOEε4 allele is associated with reduced cerebrospinal fluid levels of A 42. Neurology. 2004; 62(11): Tapiola T, Pirttila T, Mehta PD, Alafuzofff I, Lehtovirta M, Soininen H. Relationship between apoe genotype and CSF -amyloid 1-42 and tau in patients with probable and definite Alzheimer s disease. Neurobiol Aging. 2000;21(5): Andersson C, Blennow K, Johansson SE, et al. Differential CSF biomarker levels in APOE-ε4 positive and negative patients with memory impairment. Dement Geriatr Cogn Disord. 2007;23(2):