Posted 7 November 2003

This article summarizes the research findings and intellectual input of the Antecedent Biomarkers Group (ABG; see Appendix A) - the invited presenters and attendees of a workshop sponsored by the Washington University Alzheimer's Disease Research Center in St. Louis, Missouri, on May 23, 2003. This article was written by Gila Z. Reckess, MSc, under contract from the Organizing Committee. It was modified through feedback from numerous members of the ABG and, as such, represents a consensus report.

1. Introduction
The imperative to develop effective therapies or preventions for Alzheimer's disease (AD), by far the most common dementing illness of late life, increases with the ever-growing number of older adults. Diagnosis of AD remains in the hands of clinicians; there is no current test or procedure that is diagnostic. Not surprisingly, AD remains underdiagnosed and undertreated. Moreover, the clinical symptoms required for diagnosis appear to develop only after substantial cell loss has occurred in vulnerable brain regions,(1) which results after years or even decades of the underlying disease process.(2)

Current therapies are initiated only after diagnosis; their modest benefit, in part, may be explained by the fact that some irreversible brain damage already has occurred by the time dementia is recognized. The development of valid and reliable biomarkers for AD not only will aid clinicians in recognizing the disease in its earliest symptomatic stages, but also may help identify the illness before dementia or other symptoms appear. The detection of preclinical AD will be especially important should effective disease-modifying therapies be developed to allow optimal intervention (i.e., before substantial neuropathological damage has occurred and before the manifestation of dementia).

To examine the nascent field of antecedent biomarkers for AD, a group of scientists and physicians (the Antecedent Biomarkers Group) convened in May 2003 in St. Louis, Missouri, to review recent findings on the use and limitations of candidate biomarkers for AD that may have utility in preclinical stages. The group also discussed future steps for developing, testing, and refining biomarker candidates. The following is a review of the proceedings.

1.1 Definitions
Biological markers can reflect a variety of disease characteristics, including the level of exposure to an environmental or genetic trigger, an element of the disease process itself, an intermediate stage between exposure and disease onset, or an independent factor associated with the disease state but not causative of pathogenesis. Depending on the specific characteristic, biomarkers can be used to identify the risk of developing an illness (antecedent biomarkers), aid in identifying disease (diagnostic biomarkers), or predict future disease course, including response to therapy (prognostic biomarkers).

Proposed criteria for effective biomarkers in AD have been described:

"The ideal biomarker for AD should detect a fundamental feature of neuropathology and be validated in neuropathologically confirmed cases; it should have a diagnostic sensitivity >80 percent for detecting AD and a specificity of >80 percent for distinguishing other dementias; it should be reliable, reproducible, noninvasive, simple to perform, and inexpensive. Recommended steps to establish a biomarker include confirmation by at least two independent studies conducted by qualified investigators with the results published in peer-reviewed journals.(3) It would be especially useful if the biomarker could also capture the beneficial effect of disease-modifying therapy.(4)"

Ideally, an antecedent biomarker or group of biomarkers would have the above characteristics, but also would incorporate one key additional feature: the ability to detect preclinical pathology. Because AD pathogenesis begins years before onset of symptoms (and even then may reflect a late stage in the disease process, preceded by induction and latency phases of as-yet unknown duration), comprehensive screening and diagnosis require presymptomatic insight. Optimally, biomarker assays should be quick, easy, and inexpensive; safe and acceptable both to patients and physicians; and have established sensitivity, specificity, and predictive values.(5)

1.2 A case for antecedent biomarkers
Though the length of the induction period is still unknown, extensive data suggest that the underlying disease pathology of AD begins many years before clinical symptoms appear. For example, one series found that 28 percent of nondemented individuals had sufficient diffuse and neuritic plaques for postmortem diagnosis of AD,(6) despite the absence of clinical or cognitive indications of disease onset.(7) The authors speculated that these individuals remained free of cognitive impairment or decline until the AD neuropathological process led to sufficient synaptic and neuronal dysfunction and/or loss to result in the appearance of dementia.(8) These observations are consistent with a human model of preclinical AD: Plaques characteristic of AD begin to form in early adulthood in patients with Down's syndrome, but expression of dementia is delayed for decades.(9)

In light of the early, insidious pathogenesis of AD, combined with the theory that neuronal degeneration is easier to slow or halt than it is to reverse, it is vital to identify biomarkers that can detect the disease well before the development of symptoms and irreversible pathologic damage. If there were an inexpensive and effective preventative treatment that could safely be administered to the general population, the need to identify preclinical biomarkers would be less urgent. However, no such treatment for AD exists. Therefore, preclinical biomarkers could be used not only for screening and diagnosis, but also to identify target populations for new drugs. They also could serve as targets for novel prophylactic therapeutics and perhaps as endpoints for testing efficacy of medications under development.

1.3 Methodological challenges
The pursuit of antecedent biomarkers raises several methodological challenges. Though highly accurate, clinical diagnosis is only possible for symptomatic patients and can lack the specificity needed to distinguish AD from other forms of dementia. Moreover, autopsy remains the only definitive way to confirm the clinical diagnosis of AD. Without a way to diagnose presymptomatic disease, there is no gold standard with which to validate experimental findings without performing lengthy longitudinal studies, potentially extending to clinical diagnosis and eventual confirmation by autopsy. Identifying antecedent biomarkers of AD is, therefore, a formidable task.(10)

Biomarkers are further constrained by issues such as affordability and ease of measurement, particularly if used to screen large portions of the general population. In addition, they not only must be safe and acceptable to patients and clinical personnel, but also should be comprehensively characterized such that the precise sensitivity, specificity, and predictive values are known.

Moreover, before using a biomarker or battery of biomarkers to screen subsets of the general population, it is preferable from an ethical standpoint to have potential prevention and treatment options to offer individuals who test positive. At the moment, no such treatments exist, though several are in development. Continuing to explore the early pathologic stages of AD and establish antecedent markers of disease onset and progression could help in the pursuit of medical treatment by identifying target treatment populations and biological targets for treatment.

1.4 May 2003 conference
Because of the logistical challenges, research on antecedent biomarkers of AD is scarce. Reviewing candidate biomarkers of symptomatic AD and of biomarkers in other clinical fields may, therefore, provide insight into the ideal methods of determining preclinical candidates.

In the following account, the advantages and limitations of potential post- and preclinical biomarkers for AD are outlined, as discussed at the May 2003 antecedent biomarker conference. Candidate approaches include genomics and proteomics, body fluid analysis (including tau, Aβ, isoprostanes, and sulfatide), and structural and metabolic brain imaging.

As a follow-up to this day-long symposium, it is proposed that future collaborative dialogues address four key issues: continuing to explore potential candidates and limitations in neuroimaging, fluid measurements, genomics, and proteomics; reaching a consensus on the clinical distinction between healthy individuals and those with dementia of the Alzheimer's type; establishing standardized methods for sample collection and storage; and examining other fields already in the process of searching for antecedent biomarkers of preclinical pathogenesis.

2. Pathology of AD
Cognitive performance remains relatively stable in healthy individuals as they age, but rapidly declines in individuals diagnosed with the earliest stage of dementia, mild cognitive impairment (MCI).(11) A large proportion of individuals with MCI later are diagnosed with AD, and all individuals with AD first exhibit mild dementia symptoms characteristic of MCI.(12) Significant evidence suggests that by the time an individual has mild dementia due to AD, the brain already is loaded with markers of disease, including irreversible neuronal loss.(13)

Pathological diagnosis of AD remains the only way to definitively confirm the disease, and relies upon the density and distribution of senile plaques and neurofibrillary tangles (NFTs) in the brain.(14) The primary component of each represents abnormal deposition of neuronal proteins—amyloid-β (Aβ) and tau, respectively—and are rational therapeutic targets for AD. Genetic, biochemical, and animal model studies strongly suggest that the abnormal conformation and accumulation of Aβ is a critical upstream event in the pathogenesis of AD, but the exact sequence and downstream events in AD pathogenesis are not resolved.

Senile plaques, which can be neuritic (containing Aβ fibrils and dystrophic neurites) or diffuse (nonfibrillar Aβ), result from the accumulation of Aβ in the extracellular environment of the brain parenchyma. Aβ is a 38-43-residue protein (15) formed via cleavage of amyloid precursor protein (APP) at two sites, known as the β- and γ-secretase sites.(16) The two most prevalent forms in vivo are Aβ40 and Aβ42. Conversion of Aβ from soluble to insoluble forms may be a critical step in the dangerous accumulation of this peptide, as is a balance between the production and clearance of all forms of Aβ.(17)

Aβ40 and Aβ42 are both located in plaques, though high levels of Aβ42 can drive plaque formation. There are high levels of insoluble Aβ42 in cortex of affected areas in AD, with plaque formation associated with up to a 300-fold increase in insoluble Aβ42.(18) Extensive neuritic and diffuse plaques in neocortex and limbic structures are evident not only in patients with MCI, but also in about 27 percent of nondemented individuals by their mid-70s, suggesting that plaque accumulation may delineate a preclinical population.(19) Moreover, individuals with Down's syndrome, most of whom develop dementia of the Alzheimer's type at about 50 years old, begin developing plaques up to 30 years before onset of dementia, often in their teenage years.(20)

The second main component of AD pathology is accumulation of NFTs. Tau, a protein primarily expressed in neuronal axons, binds tubules and promotes polymerization to form microtubules. Though the mechanisms that lead to tau accumulation in the form of NFTs have not been confirmed, the process may be triggered when hyperphosphorylation of tau disengages tau from microtubules. Aggregated pools of unbound tau accumulate as neuropil threads in axons and as neurofibrillary tangles in cell bodies. As a result, processes degenerate, neurons die, and the threads and tangles are released into the extracellular environment.(21)

NFTs appear to represent markers of neuronal injury and death. In addition, they are not unique to AD: They also are found in nondemented individuals, particularly in certain brain regions, and increase exponentially with age.(22)

Cognitively normal individuals who have plaque accumulation indicative of preclinical AD also appear to have more NFTs and an increased rate of tangle development, suggesting an interaction between plaque and tangle formation.(23)

Though not as well-established as the connection between AD and either tau or Aβ, α-synuclein (αS), the main component of Lewy bodies, often is found in both familial and sporadic AD(24) primarily in the amygdaloid complex. It is still unclear whether coexistence of αS and AD pathology reflects an interactive relationship between the two. However, recent evidence does suggest an interaction between αS and tau, in which the two proteins promote dangerous fibrillization of each other.(25)

3. Antecedent candidates
Because AD in particular may have a long presymptomatic phase and potentially multiple underlying molecular mechanisms, and because therapeutics are under development for several pathological stages and with a variety of goals in mind, it is wise to simultaneously investigate the efficacy of multiple biomarkers.(26)

Moreover, any one given biomarker may not have sufficient sensitivity to detect all cases of the disease or adequate specificity to distinguish one pathological fingerprint from other disease profiles. Therefore, a battery of biomarkers may provide superior accuracy. For example, an effective diagnostic protocol could be comprised of two sets of measurements, one with high sensitivity to detect individuals at risk followed by a marker(s) with accurate specificity to weed out false positives.(27)

It also is essential to shift from searching only for binary markers (e.g., is a gene present or not?) to examining patterns of data.

Several types of biomarkers were discussed at the conference, each of which could potentially contribute to a battery of measurements used to diagnose preclinical AD. Candidates fell into one of four categories: -omics, including genomics and proteomics; body fluid analyses of tau, Aβ or both; neuroimaging; and isoprostanes. Preliminary findings with sulfatide were also discussed.

3.1 -omics
To date, four genes have been identified that increase the risk of AD.

The earliest, most severe cases of AD typically are caused by missense mutations in presenilin 1 (PS1), found on chromosome 14. Mutations of presenilin 2 (PS2) on chromosome 1 also cause early onset AD, though many such cases have later onset than those caused by PS1. Mutations in amyloid precursor protein (APP) surrounding the β- and γ-secretase sites are another rare cause of early onset familial AD, and overexpression of APP on chromosome 21 is thought to be responsible for AD pathology in Down's syndrome.(28) Mutations in PS1, PS2 and APP that lead to AD all result in increased Aβ42 production.

Mutations in these three genes are rare, as is early-onset AD. To date, no deterministic genes have been identified for the more common, sporadic form of AD. However, the ε4 allele of apolipoprotein E (ApoE) on chromosome 19 increases and the ε2 allele decreases susceptibility to AD. Disease risk increases with the number of ε4 alleles,(29) though some individuals with sporadic AD do not have any copies of the allele and some individuals homozygous for ε4 never develop AD.(30)

In regard to biomarkers, a preliminary study suggests a dose-dependent effect of ε4 on the amount of Aβ1-42 in cerebrospinal fluid (CSF) in subjects with a family history of AD and a mean age in their late 50s:(31) Cognitively normal individuals with a family history of AD who have no ε4 alleles have the highest amounts of CSF Aβ1-42, followed by individuals with one allele, and then by individuals with two copies of ε4. This decrease in CSF Aβ suggests that plaques may be forming in the brain, somehow resulting in a lowering of CSF Aβ, but ongoing follow-up of this population will be necessary to see if the finding is persistent and associated with a greater incidence of clinical AD over time.

The utility of ε4 alone as a definitive, predictive biomarker is currently limited, but it may have value as a member of a larger arsenal of measurements. It also continues to be useful in designing study paradigms of early or preclinical AD. Studies continue to search for other useful genetic indicators of sporadic AD.

For the comparatively small population of early-onset AD, gene sequencing of PS1, PS2 and APP could be used as both antecedent and postclinical biomarkers. However, until a preventative or curative treatment is developed, early diagnosis may raise ethical issues.

3.1.1 Lessons from cancer
Several other clinical fields already have applied genetic techniques to the search for antecedent and prognostic biomarkers and may be informative resources. The ultimate goal of these approaches is to identify predictive patterns in genetic profiles so that treatment can be tailored for each individual patient.

Cancer, in particular, often is at the forefront of genetic research, in part due to the necessity and desirability of excising the tissue of interest (e.g., the tumor). Though tissue collection is not the most realistic and viable option for neurodegenerative diseases, the techniques and experience gleaned from oncological research offer a valuable methodological perspective.

Using gene-chip microarrays, for example, macrophage inhibitory cytokine I was identified as a potential serum biomarker for prostate cancer (32). However, it often is difficult or inappropriate to focus on just one particular candidate gene. Pattern recognition approaches may have greater appeal and applicability for complex diseases such as AD because they examine a range of candidate genes as a potential group profile as opposed to a single, binary test.(33)

There are several recent examples of the prognostic ability of such tumor profiling. For example, combining the use of gene chips and cluster analysis, a group of 70 genes was identified that, when used together, predict metastasis with 83 percent accuracy in breast cancer.(34) Multiple predictive signatures also were identified for metastasis of B-cell lymphoma,(35) and one study made the remarkable discovery that a fingerprint incorporating 17 genes common to a wide range of metastasized primary tumors is capable of predicting metastasis in a wide range of cancers, including breast and lung cancer.(36)

Several oncological research areas incorporate analysis of fluids rather than tissue, a potentially more relevant approach for AD. For example, polymerase chain reaction assays have been used to examine tumor DNA in serum. With this method, methylated adenomatous polyposis coli (APC) promoter DNA was found in serum and/or plasma from 47 percent of patients with lung cancer whose primary tumor revealed signs of methylated APC, whereas the promoter was not detected in samples from any healthy controls.(37)

Proteomics also has potential applicability to AD research. A new technique called SELDi-MS (surface enhanced laser desorption ionization-mass spectrometry) is one of a large number of proteomic techniques being applied to identify biomarkers related to human diseases. It allows a computer to isolate and recognize complex protein patterns from a serum sample. Using a mass spectrometer, a complex pattern of mass sizes is produced, with peaks representing proteins of significance. The pattern is fed into a software program that isolates predictive algorithms for evaluating future samples. The more data given to the computer, the stronger the resulting algorithm. The technique does not identify specific molecules, but the patterns it reveals may be distinct enough to help detect disease. It also has the advantage of being relatively quick and efficient and requires no more than one microliter of blood from each patient.

In the first published account using this method, a serum fingerprint for ovarian cancer was isolated which correctly identified cancer in 50 out of 50 new cases.(38)

Genetic methods such as PCR and gene chip analysis and proteomic methods like SELDi-MS also may help determine the best therapeutic approach for each patient. Their use for AD, however, has yet to be fully explored. Since the search for primary deterministic genes of sporadic AD has not been fruitful, genetic pattern recognition approaches and proteomics may be particularly insightful.

3.2 Aβ and tau

3.2.1 Amyloid β
As described in the pathology section above, aggregation of Aβ in the brain seems to begin years, if not decades, before clinical symptoms arise.(39) But postmortem, pathological detection of early plaque formation clearly is not a useful biomarker, other than for diagnostic confirmation. So the question becomes whether the presence of early Aβ deposits can be identified by measurements of parameters derived from body fluids.

Mean CSF Aβ42 is significantly lower in AD patients compared to controls, but there is a large amount of overlap between the groups.(40) CSF Aβ follows a similar pattern in the very early stages of clinical dementia, with MCI patients showing overall lower levels than controls. As in AD, there is significant overlap in CSF Aβ42 between MCI and controls.(41)

There also is preliminary evidence that individuals with preclinical AD may already exhibit lower levels of CSF Aβ42.(42) Even nondemented individuals in their late 50s with at least one ε4 allele have significantly lower levels of CSF Aβ42 than individuals without ε4.(43)

Due to the significant overlap in CSF Aβ42 levels between demented and nondemented individuals, CSF Aβ and Aβ42 may not, in isolation, be useful tests for either pre- or postsymptomatic diagnosis. However, meta-analyses of current data on CSF Aβ levels may be clouded by methodological inconsistencies. Therefore, more extensive research is needed to standardize Aβ measurement methods in order to compare data from different studies and interpret overall trends.(44) Support for the diagnostic significance of plasma Aβ is less consistent than with CSF Aβ. One study suggests that plasma Aβ42 levels may predict risk of dementia.(45) Another study shows no difference between Aβ40 or Aβ42 plasma levels in AD compared with controls.(46)

However, as we begin to understand the metabolism and rate of elimination of Aβ in plasma, it may be possible to improve the utility of plasma Aβ measurements. Because the half-life of Aβ is extremely short in plasma (on the order of 10-15 minutes), it is conceivable that rapid degradation of Aβ makes it impossible to accurately measure levels that reflect what is occurring in the CNS. For example, levels of plasma Aβ do not predict plaque load in PDAPP mice, an animal model of AD. However, levels of plasma Aβ do correlate with the extent of plaque deposition in the PDAPP mouse model of AD when the monoclonal antibody m266 is administered intravenously followed by measurement of plasma Aβ. The antibody sequesters Aβ in the blood and, over a long period of time, the resulting measurements may more accurately reflect estimates of CNS Aβ levels and efflux.(47)

A third body fluid—brain interstitial fluid (ISF)—may offer new insights into Aβ metabolism. Using microdialysis for Aβ to measure ISF Aβ in PDAPP mice, it was recently shown that the half-life of ISF Aβ in the CNS is about two hours.(48) Assessment of CNS Aβ elimination rate with this method may be able to provide insight into how different molecules and the disease influence Aβ clearance.(49)

Because of its early accumulation in the preclinical AD brain, Aβ, in theory, is a prominent biomarker candidate, both for pre- and postclinical diagnostic and prognostic purposes. However, findings to date suggest that body fluid measurements are not sufficiently reliable to be used alone.

Also, confounding factors should be considered. One potential caveat, for example, is that peripheral levels of Aβ may be correlated with creatinine levels.(50) As such, these measurements are a nonspecific indicator of impaired kidney function. Peripheral levels in particular may be influenced by the general medical state of the subject; CSF levels may be less affected by such confounds.

3.2.2 Tau
Similar to trends of CSF Aβ, CSF tau also appears to distinguish controls from AD patients, with levels significantly higher in patients. However, while CSF assays for tau are far more consistent across institutions than those for Aβ, tau levels still overlap significantly between patients and controls, limiting the diagnostic usefulness of this measurement on its own.(51)

Though some studies suggest that tau is a better predictor of clinical decline than Aβ,(52) it may have less merit as an antecedent biomarker since it primarily enters and accumulates in CSF after cells begin to degenerate and die.(53) Indeed, in the same preliminary examination of nondemented 50-year-olds with a family history of AD that revealed a relationship between ε4 and CSF Aβ, no such correlation was found with total tau.(54) However, it is possible that phosphorylated tau (p-tau) levels may be more indicative, and further research is needed to differentiate the potential roles of total and p-tau in antecedent diagnosis.

3.2.3 Combining Aβ and tau
It is possible that combining a battery of biomarker measurements may improve diagnostic accuracy. For example, in a sample with 131 AD patients and 72 controls, combining measurements of Aβ and total tau in CSF yielded 92 percent sensitivity and 82 percent specificity of diagnosis,(55) suggesting that the two combined are promising in postsymptomatic diagnosis. Preclinical utility of combined Aβ and tau CSF levels has yet to be evaluated, and would require retrospective analysis of CSF samples from individuals who are asymptomatic at the time of fluid donation.

To date, 43 published studies have evaluated patterns of CSF tau in AD, 21 studies have examined Aβ, and only 16 have reported p-tau measurements.(56) Further, prospective studies of the relationships between total tau, p-tau and Aβ are needed to understand both the underlying mechanisms that trigger AD and to determine the potential combined diagnostic utility of these measures.(57)

3.3 Imaging
Structural imaging may only indirectly reflect cell loss, but it may still reveal diagnostically relevant information about disease onset and progression. While most studies to date focus on symptomatic AD, there is some emerging evidence that imaging can help predict onset of dementia and progression through the early stages of AD.

Data from magnetic resonance imaging (MRI) studies show a close link between volumetric changes and clinical symptoms. Premortem MRI of hippocampal volume is correlated with both cognitive performance and postmortem pathology,(58) as is postmortem MRI.(59)

MRI changes also appear to track cognitive decline and correlate with clinical progression. For example, the hippocampus shrinks and the temporal horn of the lateral ventricle expands in both AD and healthy individuals, but the rate of change is almost 2.5 times faster in AD patients.(60)

Moreover, evidence suggests MRI data may help predict onset of clinical symptoms and conversion from early dementia (which may be due to any of a number of pathologies) to AD. Rate of volumetric atrophy in the hippocampus, the entorhinal cortex and the ventricle all predict decline to MCI in formerly "healthy" individuals and "conversion" from MCI to AD.(61) Elevated rates of hippocampal atrophy also are associated with the presence of a single ε4 allele in healthy women in their sixth decade of life.(62) A third study found that MCI patients in the fiftieth percentile or higher for hippocampal volume had only a nine percent rate of conversion to AD within three years; those in the bottom third (lower than the first percentile) had as high as 50 percent risk of conversion.(63)

Imaging may be easier to keep uniform and consistent across institutions, unlike assay variability in Aβ measurements. Data from 38 centers around the world in a year-long, multicenter study revealed decreases in hippocampal volume and increases in temporal horn volume in AD, similar to changes already outlined in previous studies. In addition, the sample size needed for statistical power was smaller than studies using only cognitive measurements.(64)

To further enhance the effectiveness of structural imaging, it may be possible to develop nontoxic contrast agents that permeate the blood-brain barrier and produce local changes detectable by MRI. If such agents could target proteins known to differentiate healthy individuals from those with symptomatic and, ideally, asymptomatic AD, they would be useful both as antecedent biomarkers and for postclinical assessment.

One example currently under investigation is putrascine-gadolinium-Aβ (PUT-Gd-Aβ). Preliminary evidence confirms the ability of this marker to identify plaques (later histologically validated) in mice both in vivo and postmortem.(65)

Other promising tracers may work with functional and metabolic imaging methodologies, such as positron emission tomography (PET). A newly developed compound now known as PIB ([N-methyl-11C]2-[4'-(methylamino)-phenyl]6-hydroxybenzothiazole) incorporates benzothiazole-aniline derivatives that can cross the blood-brain barrier and bind to amyloid. Using PET, accumulated PIB was detected in AD patients in brain regions known to house plaque deposits postmortem, including frontal and temporal-parietal association cortices, while less PIB was found in healthy controls.(66)

IMPY ([123I/125I]IMPY, 6-iodo-2-(4'-dimethylamino-)phenyl-imidazo[1,2-a]pyridine), also has a binding affinity for Aβ in solution and in postmortem AD samples. Autoradiograms of brain slices from a mouse model of AD (Tg 2576) show that [125I]IMPY does indeed label amyloid plaques.(67) The agent's utility for human in-vivo imaging has yet to be evaluated.

3.4 Other potential biomarker candidates

3.4.1 Isoprostanes
Peroxidation of lipids results in formation of several types of isoprostanes, all of which can be measured.(68) To date, none have been evaluated for their antecedent utility, but several have apparent potential for diagnosis, prediction of disease progression and assessment of therapeutic efficacy in symptomatic dementia patients. However, due in part to assay variability or the selection criteria applied to groups of individuals studied, some inconsistencies exist among scientific teams.

In an ongoing multicenter study called Biomarkers of Oxidative Stress Study (BOSS), F2-Isoprostanes were determined to be a sensitive and reliable isoprostane biomarker of lipid peroxidation, corroborated by three assays at three different institutions.(69) Moreover, research has shown higher concentrations of these markers in postmortem AD brain tissue(70) and ventricular CSF (71) and in lumbar CSF from living patients with MCI (72) and mild AD.(73)

Though less invasive, plasma and urine analyses of F2-isoprostanes yield less consistent results than CSF or postmortem tissue: Two studies show a significant increase in patients with AD compared with controls.(74) and one found an increase in patients with MCI;(75) five studies found no difference between either patient population or controls.(76)

Because the majority of F2-isoprostanes in the periphery do not originate in the CNS, study discrepancies may reflect dilution of CNS-derived isoprostanes by those from the rest of the body, which may be influenced by confounds such as diet, exercise, body mass index, and smoking.(77) For the same reason, peripheral measurements of F2-isoprostanes may have limited utility in determining response to therapeutics.(78) Nevertheless, preliminary data suggest that CSF F2-isoprostanes may be one means to quantitatively assess response to medical and surgical interventions for AD.(79)

In a small, preliminary study, another isoprostane—8,12-iso-iPF2a-VI—was found to be present in urine at levels three times those of F2-isoprostanes and elevated in individuals with AD compared with controls but not in people with frontotemporal dementia, suggesting that this marker is specific to AD and can be measured with minute urine samples. The isoprostane also increased over the course of two years in MCI patients compared with controls.(80)

Additionally, patients with a ventricular shunt to increase CSF drainage did not decline in cognitive function but had significant reductions in 8,12-iso-iPF2a-VI and tau (but not in Aβ1-42), whereas those patients who did not receive a shunt showed significant, progressive cognitive decline over the course of one year.(81)

A major potential confound is that isoprostane levels reflect the process of lipid peroxidation, which may produce similar increases in diseases other than AD (e.g., ischemia). Because of this lack of disease specificity, isoprostanes alone may not be ideal antecedent biomarkers. However, they may have potential as a contributor to a biomarker panel, particularly when combined with measurements of Aβ and tau.(82)

Particularly in light of inter-institutional discrepancies, further investigation is necessary to determine the potential predictive utility of CSF and peripheral fluid measurements of these markers. Exchanging biological samples for analysis by a variety of institutions may help identify the source of such data discrepancies.

3.4.2 Sulfatide
Though most AD research focuses on pathology in gray matter, new evidence suggests that levels of sulfatide, a lipid found primarily in white matter, may be indicative of AD pathogenesis. By screening normal and AD brain tissue with electrospray ionization mass spectrometry (ESI/MS) for lipid changes, sulfatide was identified as a potential marker of interest. Indeed, upon further investigation, sulfatide does appear to decrease 93 percent in gray matter and 58 percent in white matter in individuals with MCI compared with healthy controls. Ceramides, the presence of which indicates sulfatide degradation, more than tripled in white matter of individuals with MCI.(83) When normalized with CSF phosphatidylinositol (PI), another white-matter lipid, CSF sulfatide distinguished nondemented individuals from those with very mild dementia with a sensitivity of 90 percent and a specificity of 100 percent—better than CSF Aβ42, tau or p-tau.(84) These results are preliminary and the marker's preclinical utility has yet to be evaluated.

4. Discussion and future directions
Due to the early development and lengthy progression of AD pathology, it is critical to develop diagnostic and therapeutic approaches for preclinical disease. Several candidate biomarkers are under investigation, including those discussed in these proceedings. However, methods of preclinical and postsymptomatic diagnosis and prognosis have yet to be established. In particular, due to the methodological and logistical challenges in identifying preclinical populations to follow, few antecedent biomarkers have been extensively studied to date.

Further research is needed to explore each of the candidate post- and preclinical biomarkers discussed above. Further collaborative efforts and dialogue on methodology should focus on the following four areas: continuing to explore candidate biomarkers; standardizing definitions of healthy aging and early AD; standardizing experimental methods; and investigating conclusions already gleaned from other clinical research areas.

4.1 Exploring current and emerging biomarker candidates
Preliminary evidence points to three main categories of biomarker candidates: genetic and proteomic; imaging; and body fluid analysis. Though none appears powerful enough to serve alone as either an antecedent biomarker or a marker of symptomatic AD, each offers potential value to develop a group of biomarkers.

The genetics of familial early-onset AD, as identified thus far, do not lend themselves to being used as biomarkers of sporadic AD, though mutations in PS1, PS2 and APP are predictive of the rarer and more aggressive early-onset variation. The ApoE genotype is of some predictive value and may be useful in combination with the development of new biomarkers. Proteomics have yet to be explored in AD.

CSF Aβ and tau are promising as both post- and preclinical biomarkers, particularly when the two are combined. Less invasive measurements of these proteins, including plasma and urine analysis, may not be as consistent or powerful, but preliminary investigations into the metabolism of Aβ in particular suggest that plasma levels may in some way be useful.

Fluid analysis of other elements, such as isoprostanes and sulfatides are currently inconclusive but promising.

Finally, traditional structural and metabolic neuroimaging as well as the use of specialized contrast agents appear to predict AD onset and clinical progression. They therefore may be powerful additions to a battery of biomarkers.

Though promising, none of the above candidates has been proven sufficient in diagnosis, prognosis or prediction of therapeutic response in either symptomatic or asymptomatic AD. Clearly, further research is necessary to decipher the role of each candidate. In order to determine ideal preclinical candidates, extensive prospective studies are essential and may require collaboration of subject pools and resources.

4.2 Standardizing definitions
To assess preclinical AD, an accepted definition of very early AD and methods to differentiate such patients from healthy controls must be agreed upon. Evidence suggests that true clinical detection of very mild dementia (including some cases of MCI) is already associated with substantial AD pathology.(85) Therefore, it is proposed that even mildly cognitively impaired individuals be distinguished from those who truly are cognitively healthy to avoid contamination of "normal aging" data with results from individuals with potential early AD.

Moreover, information to be collected in biomarker studies should be standardized across institutions and incorporate a common questionnaire that addresses potential confounds such as medications, kidney function and potentially relevant lifestyle factors. Approaches to follow-up with the "worried well" would also be beneficial in tracking the initial stages of clinical AD.

Incorporation of younger participants to extend the reach of longitudinal investigations is also encouraged. Such endeavors may involve collaboration with other academic and research organizations as evaluation of younger populations reaches beyond the typical scope of an NIH-sponsored Alzheimer's Disease Research Center (ADRC).

4.3 Standardizing methodologies
Data acquisition, sample storage and analysis methods vary greatly. Such discrepancies can be deleterious for statistical strength and consistency, as seen in the analysis of CSF Aβ.

Study design, implementation, coordination and analysis all require extensive technical, administrative and scientific knowledge and congruity, both inter- and intra-laboratory. The types of samples (e.g., neuroimages, CSF, blood samples, etc.) and methods of collection and analysis must be consistent. Once standardized, inter-laboratory collaboration would be more feasible and should be encouraged. Moreover, samples should be stored for future analysis once new tools become available, but careful consideration should be given to storage protocol.

4.4 Learning from other fields
Oncology offers many insights into the clinical applications of genetics and proteomics. Cancer, however, does not present many of the same logistical challenges as AD, in which tissue biopsy is rarely possible and pathology begins decades before symptoms. In contrast, similar methodological, theoretical and ethical challenges exist in other neurodegenerative diseases, particularly in Huntington's disease.(86) Examining ways that such fields have addressed these issues may provide valuable information for similar pursuits in AD.

5. Conclusion
Despite the urgency of identifying ways to diagnose AD before symptoms and cell loss begin, few candidates have been explored for their preclinical value. The paucity of antecedent biomarker studies is in part due to the methodological and logistical challenges of measuring the progression of a disease with a long, asymptomatic pathogenesis. The handful of biomarkers being explored for use in diagnosis and prognosis of symptomatic AD all have potential use preclinically, and combining several such measurements may provide further utility.

Having reviewed the advantages and limitations of the main biomarker candidates currently under investigation and the potential role of each as an antecedent biomarker, it now is necessary to address experimental inconsistencies and challenges and to design longitudinal, prospective studies to further evaluate potential preclinical measurements.

To that end, plans to reconvene at a second conference, combining input both from academicians and industry, are underway.

Contact:
John C. Morris, M.D.
Chair, Antecedent Biomarkers Group
Friedman Distinguished Professor of Neurology
Principal Investigator, Alzheimer's Disease Research Center
Washington University School of Medicine
4488 Forest Park Avenue, Suite 130
St. Louis, MO 63108
(314) 286-2881; Fax (314) 286-2763
morrisj@abraxas.wustl.edu or adrcedu@abraxas.wustl.edu http://alzheimer.wustl.edu/adrc2/

Notes
1. Price et al, Arch Neurol, 2001
2. Price & Morris, Ann Neurol, 1999; Rumble et al, New Engl J Med, 1989; Goldman et al, Neurology, 2001
3. Growden et al, Neuro Biol Aging, 1998
4. Frank et al, Neuro Biol Aging, 2003
5. Mayeux, Biomarkers conference
6. Price & Morris, Ann Neurol, 1999
7. Goldman et al, Neurology, 2001
8. Price et al, Arch Neurol, 2001
9. Rumble et al, New Engl J Med 1989
10. Mayeux, Biomarkers conference
11.Storandt et al, Neurology, 2002
12. Morris, Biomarkers conference
13. Price et al, Arch Neurol, 2001
14. Morris et al, Neurology, 1991; Price et al, Neurobiol Aging, 1991
15. Morishima-Kawashima et al, Neurosci Res, 2002
16. Holtzman, Alz Dis Assoc Disord, 2003
17. Holtzman, Biomarkers conference
18. Holtzman, Biomarkers conference
19. Price & Morris, Ann Neurol, 1999
20. Rumble et al, New Engl J Med, 1989
21. Trojanowski, Biomarkers conference; Lee et al, Annu Rev Neurosci, 2001
22. Price and Morris, Ann Neurol, 1999
23. Price and Morris, Ann Neurol, 1999
24. Parkinnen et al, J Neuropathol Exp Neurol, 2003; Lippa et al, Am J Pathol, 1998; Hamilton, Brain Pathol, 2000; Arai et al, Brain Res, 2001
25. Giasson et al, Science, 2003
26. Frank, Biomarkers conference
27. Mayeux, Biomarkers conference
28. Selkoe et al, Annu Rev Genomics Hum Genet, 2002
29. Corder et al, Science, 1993
30. Selkoe et al, Annu Rev Genomics Hum Genet, 2002
31. Sunderland, Biomarkers conference
33. Welsh et al, Proc Natl Acad Sci, 2003
34. Grupe, Biomarkers conference
35. Van't Veer et al, Nat Med, 2002
36. Rosenwald et al, N Engl J Med, 2002
37. Ramaswamy et al, Nat Genet, 2003
38. Usadel et al, Cancer Res, 2002
39. Petricoin et al, Lancet, 2002
40. Rumble et al, New Engl J Med, 1989; Price & Morris, Ann Neurol, 1999
41. Sunderland et al, JAMA, 2003
42. Han et al, Ann Neurol, 2003
43. Fagan et al, Ann Neurol, 2000
44. Sunderland, manuscript under review
45. Sunderland et al, JAMA, 2003; Sunderland, Biomarkers conference
46. Mayeux et al, Ann Neurol, 1999
47. Mehta et al, Arch Neurol, 2000
48. DeMattos et al, Science, 2002
49. Cirrito et al, J Neurosci, 2003
50. Holtzman, Biomarkers conference
51. Kuller et al, in press
52. Sunderland et al, JAMA, 2003
53. Clark et al, in press
54. Trojanowski, Biomarkers conference
55. Sunderland, Biomarkers conference
56. Sunderland et al, JAMA, 2003
57. Hampel et al., J Neural Trans, submitted.
58. Frank, Biomarkers conference
59. Jack et al, Neurology, 2002
60. Bobinski et al, Neuroscience, 2000
61. Jack et al, Neurology, 1998
62. Jack et al, Neurology, 2000
63. Cohen et al, Neurology, 2001
64. Jack et al, Neurology, 1999
65. Jack, Biomarkers conference
66. Jack, Biomarkers conference
67. Klunk et al, 8th International Conference on Alzheimer's Disease and Related Disorders, 2002
68. Kung et al, Brain Res, 2002
69. Montine, Biomarkers conference
70. Montine, Biomarkers conference
71. Praticò et al, FASEB J, 1998; Reich et al, Ann J Pathol, 2001
72. Montine et al, Ann Neurol, 1998; Montine et al, Ann J Pathol, 1999
73. Praticò et al, Arch Neurol, 2002
74. Montine et al, Ann J Pathol, 1999; Praticò et al, Ann J Med, 2000; Montine et al, Arch Pathol Lab Med, 2001
75. Yao et al, Neurology, 2003
76. Praticò, Biomarkers conference
77. Montine, Biomarkers conference
78. Montine, Biomarkers conference
79. Montine, Biomarkers conference
80. Praticò, Biomarkers conference
81. Praticò, Biomarkers conference
82. Praticò, Biomarkers conference
83. Montine, Biomarkers conference; Montine et al, Arch Pathol Lab Med, 2001
84. Han et al, J Neurochem, 2002
85. Han et al, Ann Neurol, 2003
86. Price et al, Arch Neurol, 2001; Morris et al, J Mol Neurosci, 2001
87. Kayson et al, Neurology, 2002

References
Arai Y, Yamazaki M, Mori O, et al. α-synuclein-positive structures in cases with sporadic Alzheimer's disease: Morphology and its relationship to tau aggregation. Brain Res. 2001 Jan 12;888(2):287-296. Abstract

Bobinski M, de Leon MJ, Wegiel J, et al. The histological validation of postmortem magnetic resonance imaging-determined hippocampal volume in Alzheimer's disease. Neuroscience. 2000;95(3):721-5. Abstract

Cirrito JR, May PC, O'Dell MA, et al. In vivo assessment of brain interstitial fluid with microdialysis reveals plaque-associated changes in amyloid-β metabolism and half-life. J Neurosci. 2003 Oct 1;23(26):8844-53. Abstract

Clark CM, Xie S, Chittams J, et al. (in press). CSF tau and Aβ: How well do these biomarkers reflect autopsy confirmed dementia diagnosis?

Cohen RM, Small C, Lalonde F, et al. Effect of apolipoprotein E genotype on hippocampal volume loss in aging healthy women. Neurology. 2001 Dec 26;57(12):2223-8. Abstract

Coleman P, Antecedent Biomarkers Conference, 2003.

Corder EH, Saunders AM, Strittmatter WJ, et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families. Science. 1993 Aug 13;261(5123):921-3. Abstract

DeMattos RB, Bales KR, Cummins DJ, et al. Brain to plasma amyloid-β efflux: a measure of brain amyloid burden in a mouse model of Alzheimer's disease. Science. 2002 Mar 22;295(5563):2264-7. Abstract

Fagan AM, Younkin LH, Morris JC, et al. Differences in the Aβ40/Aβ42 ratio associated with cerebrospinal fluid lipoproteins as a function of apolipoprotein E genotype. Ann Neurol. 2000 Aug;48(2):201-10. Abstract

Frank R, Antecedent Biomarkers Conference, 2003.

Frank RA, Galasko D, Hampel H, et al. Biological markers for therapeutic trials in Alzheimer's disease: Proceedings of the biological markers working group; NIA initiative on neuroimaging in Alzheimer's disease. Neurobiol Aging. 2003 Jul-Aug;24(4):521-36. Review. Abstract

Giasson BI, Forman MS, Higuchi M, et al. Science. 2003 Apr 25;300(5619):636-40. Abstract

Goldman WP, Price JL, Storandt M, et al. Absence of cognitive impairment or decline in preclinical Alzheimer's disease. Neurology. 2001 Feb 13;56(3):361-7. Abstract

Growdon JH, Selkoe DJ, Roses A, et al. Application of the National Institute on Aging (NIA)-Reagan Institute criteria for the neuropathological diagnosis of Alzheimer disease. J Neuropathol Exp Neurol. 1999 Nov;58(11):1147-55. Abstract

Hamilton RL. Lewy bodies in Alzheimer's disease: A neuropathological review of 145 cases using α-synuclein immunohistochemistry. Brain Pathol. 2000 Jul;10(3):378-84. Abstract

Hampel et al., J Neural Trans, in press.

Han X, Fagan AM, Cheng H, et al. Cerebrospinal fluid sulfatide is decreased in subjects with incipient dementia. Ann Neurol. 2003 Jul;54(1):115-9. Abstract

Han X, Holtzman DM, McKeel DW Jr, et al. Substantial sulfatide deficiency and ceramide elevation in very early Alzheimer's disease: potential role in disease pathogenesis. J Neurochem. 2002 Aug;82(4):809-18. Abstract

Holtzman DM. Amyloid-β binding molecule: Potential role in the pathogenesis and treatment of Alzheimer disease. Alzheimer Dis Assoc Disord. 2003 Apr-Jun;17 Suppl 2:S66-8. Review. No abstract available. Abstract

Holtzman DM. Antecedent Biomarkers Conference, 2003.

Holtzman DM. Antecedent Biomarkers Conference, 2003.

Jack CR Jr, Dickson DW, Parisi JE, et al. Antemortem MRI findings correlate with hippocampal neuropathology in typical aging and dementia. Neurology. 2002 Mar 12;58(5):750-7. Abstract

Jack CR Jr, Petersen RC, Xu Y, et al. Rate of medial temporal lobe atrophy in typical aging and Alzheimer's disease. Neurology. 1998 Oct;51(4):993-9. Abstract

Jack CR Jr, Petersen RC, Xu Y, et al. Rates of hippocampal atrophy correlate with change in clinical status in aging and AD. Neurology. 2000 Aug 22;55(4):484-89. Abstract

Jack CR Jr, Petersen RC, Xu YC, et al. Prediction of AD with MRI-based hippocampal volume in mild cognitive impairment. Neurology. 1999 Apr 22;52(7):1397-403. Abstract

Jack CR Jr. Antecedent Biomarkers Conference, 2003.

Kayson EP, the Huntington Study Group/PHAROS Investigators. Safety and feasibility of the Prospective Huntington at Risk Observational Study (PHAROS): A preliminary report. Neurology, 58:A334, 2002.

Klunk et al. 8th International Conference on Alzheimer's Disease and Related Disorders, 2002.

Kuller LH, Mehta PD, Lopez OL, et al. β amyloid and risk of dementia in the cardiovascular health study. In prep.

Kung MP, Hou C, Zhuang ZP, et al. IMPY: an improved thioflavin-T derivative for in-vivo labeling of β-amyloid plaques. Brain Res. 2002 Nov 29;956(2):202-10. Abstract

Lee VMY, Goedert M, Trojanowski JQ. Neurodegenerative tauopathies. Annu Rev Neurosci. 2001;24:1121-59. Review. Abstract

Lippa CF, Fujiwara H, Mann DM, et al. Lewy bodies contain altered α-synuclein in brains of many familial Alzheimer's disease patients with mutations in presenilin and amyloid precursor protein genes. Am J Pathol. 1998 Nov;153(5):1365-70. Abstract

Mayeux R, Tang MX, Jacobs DM, et al. Plasma amyloid β-peptide 1-42 and incipient Alzheimer's disease. Ann Neurol. 1999 Sep;46(3):412-6. Abstract

Mehta PD, Pirttila T, Mehta SP, et al. Plasma and cerebrospinal fluid levels of amyloid β proteins 1-40 and 1-42 in Alzheimer disease. Arch Neurol. 2000 Jan;57(1):100-5. Abstract

Montine TJ, Beal MF, Cudkowicz ME, et al. Increased CSF F2-isoprostane concentration in probable AD. Neurology. 1999 Feb;52(3):562-5. Abstract

Montine TJ, Kaye JA, Montine KS, et al. Cerebrospinal fluid Aβ42, tau, and F2-isoprostane concentrations in patients with Alzheimer disease, other dementias, and in age-matched controls. Arch Pathol Lab Med. 2001 Apr;125(4):510-2. Abstract

Montine TJ, Markesbery WR, Morrow JD, Roberts LJ. Cerebrospinal fluid F2-iso-prostane levels are increased in Alzheimer's disease. Ann Neurol. 1998 Sep;44(3):410-3. Abstract

Montine TJ, Markesbery WR, Zackert W, et al. The magnitude of brain lipid peroxidation correlates with the extent of degeneration but not with density of neuritic plaques or neurofibrillary tangles or with ApoE genotype in Alzheimer's disease patients. Am J Pathol. 1999 Sep;155(3):863-8. Abstract

Montine TJ. Antecedent Biomarkers Conference, 2003.

Morishima-Kawashima M, Ihara Y. Alzheimer's disease: β-Amyloid protein and tau. J Neurosci Res. 2002 Nov 1;70(3):392-401. Review. Abstract

Morris JC, Ernesto C, Schafer K, et al. Clinical dementia rating training and reliability in multicenter studies: the Alzheimer's Disease Cooperative Study experience. Neurology. 1997 Jun;48(6):1508-10. Abstract

Morris JC, McKeel D, Storandt M, et al. Very mild Alzheimer's disease: informant-based clinical, psychometric, and pathological distinction from normal aging. Neurology. 1991 Apr;41(4):469-78. Abstract

Morris JC, Price AL. Pathologic correlates of nondemented aging, mild cognitive impairment, and early-stage Alzheimer's disease. J Mol Neurosci. 2001 Oct;17(2):101-18. Review. Abstract

Morris JC, Storandt M, Miller JP, et al. Mild cognitive impairment represents early-stage Alzheimer disease. Arch Neurol. 2001 Mar;58(3):397-405. Abstract

Parkinnen L, Soininen H, Alafuzoff I. Regional distribution of α-synuclein pathology in unimpaired aging and Alzheimer Disease. J Neuropathol Exp Neurol. 2003 Apr;62(4):363-7. Abstract

Petricoin EF, Ardekani AM, Hitt BA, et al. Use of proteomic patterns in serum to identify ovarian cancer. Lancet. 2002 Feb 16;359(9306):572-7. Abstract

Praticò, Antecedent Biomarkers Conference, 2003.

Praticò D, Clark CM, Lee VMY, et al. Increased 8,12-iso-iPF2-VI in Alzheimer's disease: Correlation of a noninvasive index of lipid peroxidation with disease severity. Ann Neurol. 2000 Nov;48(5):809-12. Abstract

Praticò D, Clark CM, Liun F, et al. Increase of Brain Oxidative Stress in Mild Cognitive Impairment: A Possible Predictor of Alzheimer Disease. Arch Neurol. 2002 Jun;59(6):972-6. Erratum in: Arch Neurol 2002 Sep;59(9):1475. Abstract

Praticò D, Delanty N. Oxidative injury in diseases of the central nervous system: focus on Alzheimer's disease. Am J Med. 2000 Nov;109(7):577-85. Review. Abstract

Praticò D, Lee VMY, Trojanowski JQ, et al. Increased F2-isoprostanes in Alzheimer's disease: evidence for enhanced lipid peroxidation in-vivo. FASEB J. 1998 Dec;12(15):1777-83. Abstract

Praticò D. In vivo measurement of the redox state. Lipids. 2001;36 Suppl:S45-7. Review. Abstract

Price JL, Davis PB, Morris JC, White DL. The distribution of tangles, plaques and related immunohistochemical markers in healthy aging and Alzheimer's disease. Neurobiol Aging. 1991 Jul-Aug;12(4):295-312. Abstract

Price JL, Ko AI, Wade MJ, et al. Neuron number in the entorhinal cortex and CA1 in preclinical Alzheimer disease. Arch Neurol. 2001 Sep;58(9):1395-402. Abstract

Price JL, Morris JC. Tangles and plaques in nondemented aging and "preclinical" Alzheimer's disease. Ann Neurol. 1999 Mar;45(3):358-68. Abstract

Ramaswamy S, Ross KN, Lander ES, Golub TR. A molecular signature of metastasis in primary solid tumors. Nat Genet. 2003 Jan;33(1):49-54. Epub 2002 Dec 09. Abstract

Reich EE, Markesbery WR, Roberts LJ II, et al. Brain regional quantification of F-ring and D-/E-ring isoprostanes and neuroprostanes in alzheimer's disease. Am J Pathol. 2001 Jan;158(1):293-7. Abstract

Rosenwald A, Wright G, Chan WC, et al. The use of molecular profiling to predict survival after chemotherapy for diffuse large-B-cell lymphoma. N Engl J Med. 2002 Jun 20;346(25):1937-47. Abstract

Rumble B, Retallack R, Hilbich C, et al. Amyloid Aβ protein and its precursor in Down's syndrome and Alzheimer's disease. N Engl J Med. 1989 Jun 1;320(22):1446-52. Abstract

Selkoe DJ, Podlisny MB. Deciphering the genetic basis of Alzheimer's disease. Annu Rev Genomics Hum Genet. 2002;3:67-99. Epub 2002 Apr 15. Review. Abstract

Siemers E. Antecedent Biomarkers Conference, 2003.

Storandt M, Grant EA, Miller JP, Morris JC. Rates of progression in mild cognitive impairment and early Alzheimer's disease. Neurology. 2002 Oct 8;59(7):1034-41. Abstract

Sunderland T, Linker G, Mirza N, et al. Decreased β-Amyloid1-42 and increased tau levels in cerebrospinal fluid of patients with Alzheimer disease. JAMA. 2003 Apr 23-30;289(16):2094-103. Abstract

Sunderland T. Antecedent Biomarkers Conference, 2003.

Sunderland T. Biomarkers in older controls "at risk" for dementia. Manuscript under review.

Trojanowski JQ, Antecedent Biomarkers Conference, 2003.

Usadel H, Brabender J, Danenberg KD, et al. Quantitative adenomatous polyposis coli promoter methylation analysis in tumor tissue, serum, and plasma DNA of patients with lung cancer. Cancer Res. 2002 Jan 15;62(2):371-5. Abstract

Van't Veer LJ, De Jong D. The microarray way to tailored cancer treatment. Nat Med. 2002 Jan;8(1):13-4. Abstract

Welsh JB, Sapinoso LM, Kern SG, et al. Large-scale delineation of secreted protein biomarkers overexpressed in cancer tissue and serum. Proc Natl Acad Sci U S A. 2003 Mar 18;100(6):3410-5. Epub 2003 Mar 06. Abstract

Yao Y, Zhukareva V, Sung S, Clark CM, Rokach J, Lee VM, Trojanowski JQ, Pratico D. Enhanced brain levels of 8,12-iso-iPF(2α)-VI differentiate AD from frontotemporal dementia. Neurology. 2003 Aug 26;61(4):475-478. Abstract

Appendix A: Members and Their Roles in the Antecedent Biomarkers Group