Journal of Molecular Neuroscience

, Volume 33, Issue 1, pp 114–119

F2-Isoprostanes as Biomarkers of Late-onset Alzheimer’s Disease


    • Department of PathologyUniversity of Washington
    • Department of NeurologyOregon Health and Sciences University
  • Joseph Quinn
    • Department of NeurologyOregon Health and Sciences University
  • Jeffrey Kaye
    • Department of NeurologyOregon Health and Sciences University
  • Jason D. Morrow
    • Department of MedicineVanderbilt University

DOI: 10.1007/s12031-007-0044-1

Cite this article as:
Montine, T.J., Quinn, J., Kaye, J. et al. J Mol Neurosci (2007) 33: 114. doi:10.1007/s12031-007-0044-1


Alzheimer’s disease (AD) is a syndrome caused by a few uncommon mutations that lead to early-onset disease, occurs in adults with Down’s syndrome, but is by far most commonly seen as a late-onset disease with multiple risk factors but no causative factors yet identified. Emerging data suggests a chronic disease model for AD with latency, prodrome, and dementia stages together lasting decades. Free radical damage to lipids in brain is one pathogenic process of AD that may be quantified with F2-isoprostanes (IsoPs). Whereas brain and cerebrospinal fluid (CSF) F2-IsoPs are reproducibly elevated in AD patients at both dementia and prodromal stages of disease, plasma and urine F2-IsoPs are not reproducibly increased in AD patients. CSF F2-IsoPs may be used to assist in diagnosis and aid in objective assessment of disease progression and response to therapeutics in patients with AD.


F2-IsoprostanesAlzheimer’s diseaseLatencyProdromeDementia

Alzheimer’s Syndrome, Disease, and Dementia

What is commonly referred to as Alzheimer’s disease (AD) really is a syndrome, a common clinicopathological entity that has multiple causes. Unusual early-onset forms of this syndrome are caused by highly penetrant autosomal dominant mutations in one of three different genes: amyloid precursor protein gene, presenilin (PS) 1 gene, or PS 2 gene (Tsuang and Bird 2002). In addition, apparently the same pathological processes very commonly afflict adults with trisomy 21 or Down’s syndrome. However, it is late-onset AD (LOAD) that represents a significant and growing public health burden, a “silent epidemic,” currently affecting between 2.5 and 4 million people in the USA, and more than 10 million people worldwide (Evans 1990). The causes of LOAD are not yet clarified, but several environmental and genetic risk factors have been identified; the most potent of these being inheritance of the ɛ4 allele of the apolipoprotein E gene (Strittmatter 1995). LOAD is projected to grow to staggering prevalence in the next generation with an estimated 8 to 12 million patients by the year 2050 in the USA alone. In addition to the untold suffering by patients and their families, AD is the third most costly medical condition in the USA (Welch et al. 1992; Ernst and Hay 1994; McCormick et al. 2001). As the number of patients afflicted continues to mount, the need for safe and effective therapy to delay or avert LOAD will become imperative (Brookmeyer et al. 1998).

It is now well recognized that the pathologic processes of AD precede by as much as two or three decades clinically diagnosed dementia. Indeed, as early as 1976, Katzman applied the chronic disease model to AD and proposed the existence of a latent stage where some structural damage accrues but there are no functional or behavioral changes, followed by a prodromal stage during which more structural damage accrues and mild functional and behavioral changes occur, and ultimately by a clinical stage with substantial irreversible damage and behavioral abnormalities (Katzman 1976). AD latency and prodrome are receiving increasing attention because it is here that interventions would have the greatest public health impact and because of the growing realization that treatment strategies at these earlier stages may differ from treatment of established dementia (Martin et al. 2002).

Several clinical, neuroimaging, and clinicopathologic studies have investigated the prodromal stage of AD (reviewed in Petersen et al. 2001). In addition, clinicopathologic studies have shown that AD-type neurodegenerative changes, viz. neuritic plaques (NPs) and neurofibrillary tangles (NFTs), are commonly present in older individuals rigorously demonstrated to be cognitively normal (Price et al. 1991; Arriagada et al. 1992; Berg et al. 1993; Crystal et al. 1993; Hulette et al. 1998; Davis et al. 1999; Haroutunian et al. 1999; Price and Morris 1999; Green et al. 2000; Schmitt et al. 2000; Xuereb et al. 2000; Morris and Price 2001; Riley et al. 2002). Broadly summarizing these many studies of cognitively normal individuals >65 years old, approximately 80% or more will have NFTs in entorhinal cortex and >50% will have NPs in isocortex as identified by standard silver-stain techniques. These data from cognitively normal older individuals, along with elegant clinical and neuroimaging studies, form the basis for the widely-held view that AD has latent and prodromal phases during which damage accrues but the person does not meet clinical criteria for dementia. Progression to Alzheimer’s dementia is marked by increased density and more extensive distribution of these histopathologic features.

A very large number of pathological and experimental studies have proposed that formation of NPs and NFTs lies distal in a complex molecular cascade that remains to be fully characterized (Hardy and Selkoe 2002). Important elements in this cascade include increased oxidative damage, loss of synapses, and the formation of nonsoluble protein aggregates. Combined with the results of the abovementioned autopsy studies, these data strongly suggest that a substantial subset of older individuals who do not meet clinical criteria for dementia are experiencing not only histopathologic but also other structural and biochemical features of AD-type neurodegeneration. Importantly, this concept of prodromal AD implies that the opportunity may exist to intervene before the onset of clinically apparent dementia. Indeed, one interpretation of the apparent discordance between epidemiologic studies that associate protection from subsequent dementia by a particular drug and clinical trials of the same drug that show no therapeutic effect in patients with established dementia is that these drugs may be efficacious only during AD latency and prodrome but not during dementia (Martin et al. 2002).

Whereas LOAD is thought to contribute to approximately 70% of dementia in the elderly, the situation is often complicated by coexisting pathologic changes in about one third to one half of these demented patients. One such coexisting condition is Lewy body (LB) accumulation in limbic and isocortical brain regions; this condition is referred to as dementia with LBs (Galasko et al. 1996). Likewise, a substantial subset of demented patients with AD-type neurodegeneration also display significant vascular damage in the form of microvascular lesions (MVLs), an area under active investigation by several groups (Petrovitch et al. 2000; Xuereb et al. 2000). In further analogy to the histopathologic lesions of AD, low levels of LBs and MVLs are found in subsets of aged individuals who are not demented, perhaps indicating latency and prodromal stages for these processes as well (Davis et al. 1999; White et al. 2002). Leaving aside the most obvious limitations of transgenic mouse models of AD, there are no models of dementia that even come close to approximating the complexity of dementia in the elderly, thus underscoring the importance of studying patients that have been carefully evaluated both clinically and pathologically.

Free Radical-mediated Damage in AD

Abundant in vitro and in vivo data have strongly implicated free radical-mediated injury to diseased regions of brain as a pathogenic mechanism in AD (Montine et al. 2002a). Several potentially overlapping sources for this AD-related increased free radical damage exist including oligomers or higher order aggregates of amyloid β (Aβ) peptides, mitochondrial dysfunction, innate immune activation, excitotoxicity and others. Because of the high concentration of polyunsaturated fatty acids in brain relative to other organs, lipid peroxidation is one of the major outcomes of free radical-mediated injury to brain (Montine et al. 2002a). A critically important aspect of lipid peroxidation is that it is self-propagating and will proceed until substrate is consumed or termination occurs. In this way, lipid peroxidation is fundamentally different from other forms of free radical injury in that it is a self-sustaining process capable of extensive tissue damage (Porter et al. 1995). There are many products generated by lipid peroxidation; one group of molecules, the F2-isoprostanes (IsoPs) have received considerable attention because of their chemical stability and minimal metabolism in situ, making them ideal candidates for quantitative biomarkers of lipid peroxidation (Morrow et al. 1990).

Assays for F2-IsoPs

F2-IsoPs are formed in vivo by free radical-mediated attack on arachidonic acid followed by oxygen insertion and cyclization. F2-IsoPs are a complex mixture of 64 enantiomers contained within four regioisomeric families. In the study of AD, F2-IsoPs have been quantified by one of four different methods: commercially available enzyme-linked immunosorbent assays (Feillet-Coudray et al. 1999), two different gas chromatography–mass spectrometry (GC-MS) stable isotope dilution methods that we refer to as the original method (Morrow et al. 1990) and modified method (Pratico et al. 1998a), and most recently by liquid chromatography (LC)-MS (Bohnstedt et al. 2003). The two GC-MS methods are similar and quantify subsets of F2-IsoPs that coelute with the deuterated internal standard used; this will be key to comparing studies presented in detail below. The original GC-MS method uses a commercially available deuterated F2-IsoP, \( 8{\text{ - }}iso{\text{ - PGF}}_{{{\text{2 $ \alpha $ }}}} \) (also known as \( {\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - III}} \)), as internal standard and quantifies those F2-IsoPs in the peak that comigrate with this molecule (Morrow et al. 1994). Morrow and colleagues have shown by reverse phase high-performance liquid chromatography and electrospray ionization MS that this peak not only contains \( 8{\text{ - }}iso{\text{ - PGF}}_{{{\text{2 $ \alpha $ }}}} \) but also additional as yet uncharacterized F2-IsoPs (Morrow et al. 1994); it is for this reason that the subset quantified by the original GC-MS method is conservatively referred to as “F2-IsoPs.” The modified GC-MS method uses a different GC protocol and additional internal standards, \({\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - III,}}\)\({\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI,}}\) or \( {\text{8 - ,12 - }}iso{\text{ - iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI}} \) (Pratico et al. 1998a, 2000). This assay quantifies the peak that comigrates with each deuterated standard and refers to what is quantified as \({\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - III,}}\)\({\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI,}}\) or \({\text{8 - ,12 - }}iso{\text{ - iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI}}{\text{.}}\) Recently, we showed that \( {\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI}} \) comigrates with \( {\text{8 - }}iso{\text{ - PGF}}_{{{\text{2 $ \alpha $ }}}} \) in the original GC-MS method, proving that \( {\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI}} \) is included in the peak we call F2-IsoPs, along with \( {\text{8 - }}iso{\text{PGF}}_{{{\text{2 $ \alpha $ }}}} {\left( {{\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - III}}} \right)} \) and other as yet uncharacterized F2-IsoPs. In contrast, we found that the other isomers, deuterated \({\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - IV}}\) or \({\text{8 - ,12 - }}iso{\text{ - iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI,}}\) do not comigrate with the F2-IsoP peak quantified by the original GC-MS method.

F2-IsoPs in Brain

The first step in investigating lipid peroxidation in LOAD was to establish relevance in human postmortem tissue. The main advantage of this approach is that autopsy classification remains the gold standard for diagnosing not only LOAD but also common comorbid conditions such as LB disease and significant MVLs. A disadvantage is that almost all patients have advanced disease at the time of death. This point is particularly important because it means that finding changes postmortem does not inform about whether these processes occur early in the disease process, making them potential therapeutic targets or late-stage consequences of the disease.

Two groups have measured F2-IsoPs in brain of patients who died with advanced LOAD. Pratico et al. (1998b), demonstrated elevated F2-IsoPs that comigrated with \( {\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - III}} \) and \( {\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI}} \) levels in frontal and temporal lobes of AD patients compared to controls using the modified GC-MS method. We expanded these findings by measuring F2-IsoPs (that contain \( {\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - III}} \) and \( {\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI}} \) among others) in temporal and parietal cortex, hippocampus, and cerebellum of AD patients and age-matched controls, all with short postmortem intervals using the original GC-MS method (Reich et al. 2001). Our data also showed greater lipid peroxidation in diseased regions of AD brain but not in cerebellum, a region relatively spared by LOAD.

F2-IsoPs in Ventricular CSF

Cerebrospinal fluid (CSF) obtained from the lateral ventricles at autopsy has also been assayed for F2-IsoPs. These studies represent a bridge between postmortem tissue analysis that still permits accurate diagnosis of LOAD and comorbid processes, and the analysis of CSF from the lumbar cistern of living patients where processes contributing to dementia can be less accurately diagnosed but analysis can be carried out earlier in the course of disease. We and others have determined the concentration of F2-IsoPs in ventricular CSF obtained postmortem and found elevations in LOAD patients compared to age-matched controls (Montine et al. 1998; Pratico et al. 1998b) (Table 1). Although different deuterated standards were used and slightly different were results achieved, both methods had similar control values and both showed a statistically significant increase in ventricular CSF F2-IsoPs in AD patients compared to controls. Our group went on to demonstrate that ventricular CSF F2-IsoP concentrations in AD patients are significantly correlated with indices of neurodegeneration (Montine et al. 1999c).
Table 1

Data are the mean values from published references (noted in text)


Original GC-MS Method

Modified GC-MS Method









V-CSF (pg/ml)













L-CSF (pg/ml)






15 or 25

66 or 68


G = GC-MS method, O = original, M = modified, St = deuterated internal standard used, III = iPF2a-III (also known as \( {\text{8 - }}iso{\text{PGF}}_{{{\text{2 $ \alpha $ }}}} \)), VI = \( {\left( {{\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI}}} \right)} \), iso-VI = \({\text{8 - ,12 - }}iso{\text{ - iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI,}}\) V-CSF = ventricular CSF, L-CSF = lumbar CSF, C = control, AD Alzheimer’s disease, Δ = change in AD patients compared to age-matched controls, + = statistically significant increase, nd = no significant difference between AD and control values

F2-IsoPs in Lumbar CSF

The first study of probable AD patients early in the course of dementia showed that F2-IsoPs are significantly elevated in lumbar CSF compared to age-matched hospitalized patients without neurologic disease (Montine et al. 1999a). This study used the original GC-MS method and quantified those F2-IsoPs that comigrate with deuterated \({\text{8 - }}iso{\text{ - PGF}}_{{{\text{2 $ \alpha $ }}}} {\left( {{\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - III}}} \right)}.\) The average duration of dementia in these probable AD patients at the time of CSF examination was less than 2 years, whereas the average duration of AD is between 9 and 12 years. This same result has been observed by us in additional groups of probable AD patients and controls (Montine et al. 1999b, d, 2001). As expected, the concentration of F2-IsoPs was lower in lumbar CSF than ventricular CSF because of both a rostrocaudal gradient and likely lower levels earlier in the disease. Importantly, a different laboratory examining probable AD patients and controls using the modified GC-MS method and deuterated \( {\text{8 - ,12 - }}iso{\text{ - iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI}} \) as standard obtained similar results (Pratico et al. 2000). Although different internal standards were used in these studies, there again was good agreement for values obtained in control individuals and AD patients (Table 1). Finally, these investigators extended their studies to patients with amnestic mild cognitive impairment (MCI), a condition that appears to represent, at least in some patients, prodromal AD; individuals with MCI were reported to have \( {\text{8 - ,12 - }}iso{\text{ - iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI}} \) lumbar CSF concentrations that were intermediate between controls and patients with AD (Pratico et al. 2002).

In addition to providing mechanistic information about neurodegenerative disease pathogenesis and a means to quantitatively assess response to antioxidant therapeutics, lumbar CSF F2-IsoP levels may also provide information that is useful in diagnosis of diseases where it is elevated early. We tested the hypothesis that quantification of lumbar CSF F2-IsoPs, along with CSF Aβ42 and total tau levels, improves laboratory diagnostic accuracy for AD in patients with probable AD, dementias other than AD, and age-matched controls (Montine et al. 2001). Individuals were classified as AD or non-AD by a commercially available test measuring CSF Aβ42 and tau levels (95% sensitivity, 50% specificity), by CSF F2-IsoP and Aβ42 levels (90% sensitivity, 83% specificity), and by combined analysis using CSF F2-IsoP, Aβ42, and tau levels (84% sensitivity, 89% specificity). These results indicate that lumbar CSF F2-IsoP quantification can enhance the accuracy of the laboratory diagnosis of AD; however, this conclusion is based on a single study and these findings need to be replicated.

Another potential application of lumbar CSF F2-IsoPs is objective assessment of response to therapeutics. We pursued this question in a longitudinal assessment of lumbar CSF F2-IsoPs in a group of patients with mild probable AD followed for 1 year. Figure 1 shows the percent change in CSF F2-IsoPs observed in these 40 AD patients stratified for dietary supplementation with no antioxidant vitamins, α-tocopherol, or α-tocopherol plus ascorbate (no patients took ascorbate alone). Patients without dietary supplementation showed an approximately 50% increase in CSF F2-IsoPs over the 1-year period. Others have shown a smaller but significant longitudinal increase in CSF F2-IsoPs in patients with MCI (de Leon et al. 2006). In consonance with recently reported epidemiologic observations on the stratification of risk for AD (Zandi et al. 2004), we observed a significant pharmacologic effect only in that group that supplemented their diets with α-tocopherol and ascorbate but not α-tocopherol alone (Quinn et al. 2004).
Figure 1

Patients with mild AD who took antioxidant supplements for a 1-year period in between spinal taps that were assayed for F2-IsoPs. Patients who took no supplements or only α-tocopherol (E) (n = 20) both showed approximately 40% interval increase in CSF F2-IsoPs and are grouped together in the “No E plus C” category; no patient took only vitamin C (C). Nine patients took E plus C for the entire year and they showed a significantly suppressed interval increase in CSF F2-IsoPs (Mann–Whitney test had P < 0.01)

F2-IsoPs in Plasma and Urine

Although obtaining CSF from the lumbar cistern is not associated with significant risks, even in the elderly, when performed by experienced physicians, spinal taps can be stressful and are not easily obtained in most clinics. For these reasons, several investigators have pursued quantification of F2-IsoPs in plasma or urine. Like most data for peripheral biomarkers of neurodegenerative disease, the results have been conflicting. In large measure this is likely because of the multiple uncontrolled physiologic processes that influence the concentration of CSF-derived molecules in blood and urine and the relatively small contribution of brain-derived F2-IsoPs to the total pool of peripherally-derived F2-IsoPs in blood (Feillet-Coudray et al. 1999; Waddington et al. 1999; Montine et al. 2000, 2002b; Pratico et al. 2000, 2002; Tuppo et al. 2001; Bohnstedt et al. 2003).

The reasons for these conflicting data are not clear, but a few points are worth considering. F2-IsoPs are generated by every cell, and therefore, peripheral production unrelated to CNS disease could easily confound interpretation of blood or urine levels in AD patients. Indeed, the amount of brain-derived F2-IsoPs contributing to plasma levels is many times smaller than peripherally derived F2-IsoPs. Moreover, several factors such as diet, degree of exercise, and body mass index all appear to influence systemic F2-IsoP levels. Regardless of the reasons for the difference in peripheral F2-IsoPs in AD patients by different groups of investigators, quantification of peripheral F2-IsoPs cannot be used to objectively assess response in CNS to systemically administered therapeutics because of the lack of organ specificity for F2-IsoPs. If a new drug lowers peripheral F2-IsoPs, is that because of an antioxidant effect in liver or brain? Even if the new drug brought symptomatic relief, it would be an assumption that the mechanism was by antioxidant actions in CNS until measurements within the CNS compartment were made. Thus, whereas plasma and urine F2-IsoPs as peripheral biomarkers of AD are desirable, both technical and theoretical concerns necessitate continued use of CSF.


The clinicopathological entity referred to as AD is a syndrome that is (1) caused by a few rare genetic mutations, (2) a common outcome in adults with Down’s syndrome, and (3) by far most commonly seen as a late-onset disease with multiple risk factors but no causative factors yet identified. Emerging data suggests a chronic disease model for AD with latency, prodrome, and dementia stages together lasting decades. Free radical damage to lipids in brain is one pathogenic process of AD that may be quantified with F2-IsoPs to assist in diagnosis and aid in objective assessment of disease progression and response to therapeutics. Whereas CNS F2-IsoPs are reproducibly elevated in AD patients at both dementia and prodromal stages of disease, we conclude by a variety of analytical methods that plasma and urine F2-IsoPs, including \({\text{iPF}}_{{{\text{2 $ \alpha $ }}}} {\text{ - VI,}}\) are not reproducibly increased in AD patients. In addition, only a very small fraction of plasma F2-IsoPs derive from CSF, and so interpretation of their changes in plasma or urine is limited with respect to CNS disease or response of CNS to systemic drug exposure.


This work was supported by the Nancy and Buster Alvord Endowment and grants from the NIH (AG05136, AG23801, and AG24011).

Copyright information

© Humana Press Inc. 2007