PET Diagnosis Poised for Prime Time? FDA Wants Consensus, Better Trials

06 May 2003

Successful radiotracers for positron emission tomography (PET) diagnosis of AD are a mere year or two away, according to leading researchers who met on May 1-2 in New York. The real challenge now will be in convincing the Food and Drug Administration that the technology is valuable.

The Institute for the Study of Aging (ISOA) organized the meeting, titled "Facilitating the Development of PET Diagnostics for Alzheimer’s Disease," to evaluate the status of the technology in early diagnosis, disease monitoring, and drug development. The meeting brought together PET experts, officials from the National Institutes of Health, economists, and insurance experts. The general consensus among attendees was that PET is effective in identifying people at risk before the appearance of clinical symptoms.

The number of performed PET procedures doubled between 2001 and 2002 to an estimated 500,000, according to Nadeem Ishaque, a researcher at General Electric. The technology is sensitive to nanomolar concentration of biomolecules, can specifically target a large number of moieties, and can be quantitative. It is limited in its image quality-it has poor spatial resolution and high image noise-but those are surmountable problems, Ishaque said.

The most commonly used PET tracer, F18-labeled 2-Deoxy-2-fluoro-D-glucose (FDG), is an excellent marker of metabolic activity in the brain, according to Daniel Silverman of the University of California in Los Angeles. Even in cases that cannot be detected by standard clinical tests, brains are already hypometabolic and can be labeled as such with FDG. In a multicenter trial that used both brain imaging and autopsy data, PET had 93 percent sensitivity, 76 percent specificity, and an overall diagnostic accuracy of 86 percent. (Silverman et al., 2001: See ARF news story).

The specificity rate was lower than the sensitivity because hypometabolism can also result from other disorders, such as depression and hypothyroidism, yielding false positives for AD. Amyloid labeling may, therefore, be a more specific approach to diagnose early AD. There are several amyloid-specific radiotracers in development, three of which are the furthest ahead.

William Klunk has previously reported results with a thioflavin derivative called BTA-1. This Aβ imaging agent crosses the blood-brain barrier and binds amyloid with high specificity (see ARF related news story). At the meeting, Klunk showed dramatic images of labeling with 6-OH-BTA-1, also called the Pittsburgh compound B (PIB). In transgenic mice, the agent even identifies individual plaques (see also Bacskai, Klunk, and Engler presentations in Ipsen meeting report).

UCLA’s Jorge Barrio presented the results of labeling with F18-labeled FDDNP, which specifically labels deposits of amyloid plaques and neurofibrillary tangles (NFT) in living AD patients (see ARF related news story).

A third group, with Hank Kung at the University of Pennsylvania, is developing radiolabeled stilbenes and iodine-labeled imidazole pyrimidine compounds such as IMSB and TZDM (Kung et al., 2001; Kung et al., 2003). Any of the three projects could reach the clinic within a year or two; however, to be validated they need testing in prospective trials of AD or mild cognitive impairment that include adequate follow-up.

Kung pointed out that Aβ aggregates have two distinctive binding sites for tracers, binding to which is not mutually exclusive. He added that it is important to consider which binding sites are most significant clinically; researchers may need to use multiple ligands to get the whole picture. Using multiple probes could benefit the diagnosis of related diseases, as well. Aβ, tau, and α-synuclein all interact with each other in vivo, but various probes bind to the three elements differently. For example, iodine-labeled IMSB, but not iodine-labeled TZDM, binds to neurofibrillary tangles. The different probes could, therefore, be used to quantify the three elements and help distinguish among different diseases, which sometimes overlap in their pathologies and symptoms (see ARF related news story).

The participants agreed that no single imaging technology will be useful for every purpose. What the best approach is may depend on what one wants to find. For example, FDG-PET is best suited to detect neuronal activity, fMRI to measure cognitive reserve, and FDDNP-PET to establish plaque-tangle load.

Eric Reiman, University of Arizona, advocated using both PET and MRI in clinical trials to increase confidence in a drug’s clinical and disease-modifying effects and to minimize the chance of missing the drug’s therapeutic value. The technology can also be used to test the relevance of the amyloid hypothesis, Reiman said. To evaluate PET, researchers will need to design phase I studies-particularly longitudinal studies-that can run in parallel with other drug development. Reiman estimated that in a clinical trial, following a relatively small number of people over a relatively short time would be enough to detect an effect.

Reiman struck an optimistic note in an otherwise somber outlook on the FDA’s expectations for PET clinical trials. "It’s pretty clear that the FDA is ratcheting up to get higher and higher efficacy," said Henry Royal of the Mallinckrodt Institute of Radiology at Washington School of Medicine in St. Louis, Missouri. But imaging studies are rarely done with the same level of robustness as is required of other medical studies, Royal said. He added that "imaging people don’t think randomizing works," even though drugs are evaluated in randomized clinical trials.

To make brain imaging routine in AD diagnosis, the imaging community will either have to find a way to meet the FDA’s expectations or convince the FDA to alter its requirements. For example, even though UCLA received its first PET grant in 1978, the technology is still underutilized 25 years later, mostly because it is considered unavailable and expensive.

In the long run, many speakers argued, the cost of preventing or delaying AD may more than offset the cost of PET scans. In the US, AD ranks third in cost after cancer and cardiovascular disease; hospitalization and skilled nursing facilities represent the bulk of the expense. Studies show that available treatments can lower costs significantly. For example, donepezil treatment reduces cost by about $4,000 per patient per year (Hill et al., 2002). Other cholinesterase inhibitors such as rivastigmine and galantamine have had similar effects on cost-effectiveness.

Using imaging to augment clinical diagnosis will also result in more true positives and fewer false positives receiving inappropriate therapy, participants argued (see ARF related news story). While costs may increase from better detection and the resulting increase in early treatment, patients are likely to spend less time in more expensive disease states. Moreover, PET has been shown to be more effective than clinical criteria in evaluating early cases of AD. Because many clinicians lack expertise in psychometric tests, they fail to recognize symptoms in up to 72 percent of cases, said ISOA’s executive director, Howard Fillit. Despite all this, the American Academy of Neurology does not recommend either PET or SPECT for routine screening and, according to some attendees, neurologists rarely consider PET an option in diagnosis.

Imaging experts should be careful in claiming that imaging outperforms clinical criteria, William Jagust from the University of California in Davis cautioned. This view is based on a direct, but invalid, comparison of studies of imaging accuracy with clinical studies, which draw subjects from different populations. At the same time, if imaging experts consider clinical criteria inferior to PET, how can they use them to validate the technology? "You can’t have it both ways," Jagust said. The ideal approach combines imaging with clinical and genetic markers in the same study, some said, and current clinical trials are indeed doing so. Prevention strategies may be more effective in ApoE4 carriers, for example, because of their more rapid metabolic decline. This strategy may also become more effective as more genetic markers are identified.

A combination of those approaches should be used to attack AD in the silent period, when available interventions may do some good, said UCLA’s Barrio. Just as we don’t wait for a heart attack to prevent heart disease, he said, people should use PET and other strategies to catch AD before symptoms appear.

In the absence of effective intervention strategies, using PET-amyloid imaging raises issues of genetic testing, such as with Huntington’s disease. Researchers will have to be careful to pace preclinical diagnoses and prevention interventions so that there is something to offer those who are diagnosed with the new technology, the participants concluded.-Apoorva Mandavilli.

Apoorva Mandavilli is a science writer and editor based in New York.

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

News Citations

Paper Citations

Kung HF, Lee CW, Zhuang ZP, Kung MP, Hou C, Plössl K. Novel stilbenes as probes for amyloid plaques. J Am Chem Soc. 2001 Dec 19;123(50):12740-1. PubMed.
Kung MP, Skovronsky DM, Hou C, Zhuang ZP, Gur TL, Zhang B, Trojanowski JQ, Lee VM, Kung HF. Detection of amyloid plaques by radioligands for Abeta40 and Abeta42: potential imaging agents in Alzheimer's patients. J Mol Neurosci. 2003 Feb;20(1):15-24. PubMed.
Hill JW, Futterman R, Mastey V, Fillit H. The effect of donepezil therapy on health costs in a Medicare managed care plan. Manag Care Interface. 2002 Mar;15(3):63-70. PubMed.

External Citations

ApoE Catalyst Conference Explores Drug Development Opportunities

20 Jul 2003

On May 29 and 30, the New York City-based Institute for the Study of Aging convened a small symposium to take the pulse of current research on ApoE, the leading risk factor for nonfamilial AD. The gathered scientists were charged particularly with identifying potential areas for therapy development. The participants wrestled with the question: Which of the myriad roles described for ApoE in the decade since its original linkage to Alzheimer’s are most relevant to the pathophysiology of this disease? (Answer: Depends on whom you ask.) They also debated whether enough is known by now to shift away from open-ended basic exploration and focus instead on drug discovery in particular areas. (Answer: Ditto.) The researchers discussed how to deal with the pharmacogenomic twist to ApoE, i.e., whether the divergent effects of the known isoforms ApoE2, 3, and 4 leave open the possibility that an ApoE-based drug can be found that would help AD patients of all ApoE genotypes. (Answer: Depends on the approach, but probably not.) Finally, some scientists asked how to avoid potential cardiovascular side effects of any ApoE-based drug, given how intermeshed ApoE’s role is with lipid and cholesterol metabolism. (Answer: Too soon to tell.)

Despite the many open questions, there was a general sense that the last decade of research has produced several strategies for drug developers to test their mettle. This conference report extends a recent ApoE news update by summarizing presentations and topics not recently covered on the Alzheimer Research Forum.

ApoE’s Strong Genetics
J. Wesson (Wes) Ashford of the Stanford/VA Alzheimer’s Center in Palo Alto, California, set the stage with a general review of the effect the ApoE genotype has on the epidemiology and onset of AD. He suggested that the onset of AD, like that of mortality and most age-related diseases, follows a descriptive, fundamental aging theory developed in 1825 by the British actuary Benjamin Gompertz. It holds that the death rate of a population doubles every set number of years. For humans aging without AD, this number has evolved to be 7.5 years for women and 8.2 for men. The specific Gompertz parameters underlying AD are not understood, but even so, it is clear that ApoE genotype changes this curve, Ashford said. Despite regional and ethnic variation in its prevalence, the ApoE4 allele overall is likely responsible for half of all nonfamilial AD cases in the US. First identified as a major AD risk factor by Allen Roses and colleagues, then at Duke University, in 1993 (Corder et al., 1993) , this allele occurs in 22 percent of the population, but in 60 percent of AD cases, whereas the E2 and E3 alleles together occur in 78 percent of the population, but only 40 percent of AD cases. ApoE exerts its main effect by influencing when people get AD, Ashford said. In a clinic population, the E4/4 genotype reduces the mean age of onset to 68 from 74 in E3/3 carriers. Environmental factors and nonspecific genes also play a role in determining AD risk and may, in fact, have a relatively greater effect in very old age (see also Silverman et al., 2003). Ashford suggested that the E2 and E3 alleles, which evolved from the ancestral E4, better support the remodeling of dendrites and minimize neuronal stress over time (see also Live Discussion). E4 also has an inferior ability to handle (excess) animal fat, and its exposure to contemporary environmental conditions with its Western diet and longer lifespans could have made it a susceptibility allele for coronary heart disease and AD (see Corbo and Scacchi, 1999). Allen Roses of GlaxoSmithKline in Research Triangle Park recalled earlier attempts at studying AD in populations that have high E4/4 rates, including African pygmies, Australian aborigines and native Brazilian tribes. The study was difficult, in part because these groups tended to die too young to develop AD in significant numbers, but the few who were available for study had massive neurodegeneration.

How to Make a Drug against a Multifaceted Target?
Bruce Teter reviewed the hypothesis that AD results when regenerative processes fail. This failure accumulates as a person’s normal balance of neuritic degeneration and regeneration shifts with age, particularly in brain regions that require the highest levels of synaptic and neuritic plasticity (see Teter and Ashford, 2002)). ApoE isoforms and expression levels influence this process on several levels, including their differential effects on neuritic sprouting and on lipid transport in and out of neurons and astrocytes. A dozen studies have shown that ApoE3 promotes neuritic sprouting more than E4. In addition, Teter’s work with hippocampal slice cultures and research by other labs indicates that ApoE4 actively dampens sprouting depending on its expression level, yet how it might do so remains unclear.

Developing therapies that target ApoE is a hugely tempting goal because of its central importance in AD risk, but the protein’s contradictory effects make it complicated to attain, Teter noted. Any future ApoE-based drug must manage to suppress some of its activities, but not others. For example, a drug that simply increases ApoE expression might help E3 carriers; however, in E4 carriers it would help with those effects that result from E4 deficiency, but might also hurt if E4 indeed impedes regeneration. As it is, a person’s ApoE genotype could affect how they respond to existing and proposed drugs including statins, NSAIDs, antioxidants, and experimental ones such as cerebrolysin (see Drugs in Clinical Trials), growth factors, and their mimetics. This may occur through drug effects on ApoE expression levels, Teter said.

Teter recommended that the field focus research on the molecular mechanisms of neurite outgrowth and the regulation of human ApoE gene expression. He urged further study of the dose-dependence of important ApoE effects, as well as consistent use of physiologically relevant doses of ApoE as a way to explore which of the manifold observed ApoE effects are likely at play in disease. Finally, the general conundrum in AD drug development-prevention beats treatment, but how does one identify presymptomatic patients without a predictive biomarker?-applies particularly to ApoE, which, above all, brings on the disease early.

Working One’s Way up from ApoE’s Structure
Robert Mahley anchored a series of presentations from researchers at the Gladstone Institute of Neurological Diseases at the University of California, San Francisco. He said that the field’s poor understanding of the disease’s multifactorial nature and of the breadth of its pathogenesis was holding back drug development. Research on ApoE effects other than its interaction with Aβ needs more attention, Mahley said. Despite swift progress in recent years, the ApoE field remains short of the point where it could start excluding hypotheses and zoom in on one key, testable principle. Mahley outlined three ApoE pathways delineated by the Gladstone group and others, all of which could provide entry points for drug development. The structure of the ApoE protein provides a window to study isoform differences from the molecular level up, Mahley said. A single amino acid difference between ApoE3 and E4 at residue 112 enables an intramolecular interaction between structural domains in ApoE4, but not E3. Mahley collaborates with Karl Weisgraber, also at Gladstone, on these studies (see Weisgraber comment). That sequence difference also causes ApoE4 to be less stable and assume an intermediate structure called a molten globule. This structure is reactive in a way not seen with ApoE3; it readily insinuates itself in membranes and may explain the lysosomal leakage associated with ApoE4 that these researchers have described (see ARF related news story).

Another pathway where these structural differences play out concerns Aβ production. Mahley and colleagues use a neuronal culture assay in which the presence of ApoE4 increases Aβ production more strongly than does E3. This assay serves to test the Gladstone group’s primary therapeutic approach, namely, the discovery of small-molecule compounds that can disrupt ApoE4’s domain interaction and thus change the conformation of ApoE4 to resemble that of E3. Weisgraber’s group discovered such compounds, including GIND-25 and GIND-105, with Irvin (Tack) Kuntz at UCSF, whose lab developed DOCK, one of the first academic software packages to support the searching of compound libraries and structure-based drug optimization. The scientists are currently developing a FRET assay to enable high-throughput screening of libraries for compounds that block domain interaction.

ApoE proteolysis represents a third pathway that may be amenable to drug development, Mahley suggested. Not all ApoE gets secreted after it’s made, he said. ApoE4, much more than E3, is susceptible to cleavage by a protease, and truncated carboxyterminal fragments resulting from this reaction can fatally disrupt the cytoskeleton in vitro. Consequently, identifying and blocking the responsible protease might lead to a therapy, Mahley suggested.

Yadong Huang, also of the Gladstone Institute, continued in this vein. His lab pursues the hypothesis that neurons respond to age-related stressors, including reactive oxygen species, brain injury, or Aβ deposition, with an increase in ApoE expression that is meant for repair. Instead, the ApoE aggregates become proteolytically cleaved into fragments that cause trouble in the cytosol. Toward demonstrating the presence of such fragments, Huang first showed antibody staining near neurofibrillary tangles of human AD brain and immunoblots of human AD brain supernatant (see also (Huang et al., 2001). He noted that ApoE fragmentation does not occur in human ApoE-transgenic mice developed by Lennart Mucke, also of Gladstone, which have the transgene driven by the glial GFAP promoter, but that it does occur in similar strains of mice-developed by Gladstone’s Mahley, Robert Pitas, and Mucke-which have the transgene driven by the neuronal NSE promoter. In NSE-ApoE4 mice, in particular, the carboxyterminal-truncated ApoE fragments appear to accumulate with age, Huang said. He suggested that ApoE4 domain interaction may render this isoform susceptible to proteolysis, and reported that two ApoE4 mutants created to abolish domain interaction lost that susceptibility.

Huang presented preliminary data of his group’s effort to identify the responsible enzyme, called ApoE-cleaving enzyme (AECE). A chymotrypsin-like serine protease that primarily cleaves ApoE at positions 268 and 272, with E4 being more susceptible than E3 to the cleavage. To test for in-vivo effects of these fragments, the scientists expressed fragments of three different lengths in transgenic mice using the Thy-1 promoter. Mice expressing high levels of the longer carboxyterminal-truncated fragment ending at residue 272 (which includes ApoE’s lipid-binding domain), died at two to four months with hippocampal neurodegeneration and indications of neurofibrillary pathology. By contrast, mice expressing a shorter fragment ending at residue 240 (which lacks the lipid-binding domain) were viable and had no neurodegeneration, Huang said. Mice expressing low levels of the long fragment were viable but showed a deficit in water-maze learning and memory tests.

The AECE enzyme should be investigated as a therapeutic target, Huang suggested. The ensuing discussion focused on where those fragments truly are, how they can be definitively identified in vivo separately from full-length ApoE, and what other substrates the AECE enzyme might cut, i.e., how specific this ApoE degradation is.

Keith Crutcher, University of Cincinnati, updated his original hypothesis of ApoE neurotoxicity (see ARF Live Discussion) with further experiments on ApoE fragments, which in his hands also are toxic to cultured neurons, but appear to be generated from different sites in the protein. They are cut in ApoE’s hinge region, around residue 190, and are therefore smaller than the ones studied by the Gladstone group. Crutcher is working with a startup company to develop drug leads from heparin-derived oligosaccharides that inhibit ApoE toxicity in vitro. All agreed that knowing exactly which fragments are formed and controlling carefully the level of fragment expression in mouse strains will be important in proving their physiological relevance. The approach also hinges on the question of whether ApoE is indeed widely expressed in adult neurons in Alzheimer’s. This view has some experimental support (see e.g., Xu et al., 1999), but remains debated.

Lennart Mucke began his presentation by saying that he believes there is ample evidence to support ApoE generation in neurons, as well as in glial cells. Current work in his lab focuses on analyzing domain-interaction mouse models and ApoE fragment-expressing mice generated by the Weisgraber and Huang labs, respectively. Mucke then reviewed published work describing the human NSE-ApoE mice strains generated in his lab. Several different types of experiments in these mice corroborate Teter’s observation about the twofold isoform effect, whereby ApoE3 not only is more protective than E4, but E4 also gains a detrimental function. For example, human ApoE3-expressing mice performed well, but ApoE4-expressing mice performed particularly poorly-even worse than did mice with no ApoE at all-in water-maze experiments that require spatial learning or spatial memory. Mucke emphasized that this double-whammy is a major theme on which data from several labs are converging.

Mucke also recapped experiments on sex-specific ApoE effects (see Raber et al., 2002). Sex steroids appear to protect male mice in certain memory paradigms, and ApoE4- but not ApoE3-transgenic male mice become particularly vulnerable when the action of those steroids is blocked. ApoE4 somehow appears to reduce androgen receptor levels in the mice, and androgen treatment improved performance in ApoE4-transgenic females. Given a string of recent setbacks in trials of estrogen replacement therapy (see ARF related news story), this approach needs to be rethought, Mucke said, although there may still be subgroups of patients who benefited, but do not appear in the overall analysis.

Mucke then reviewed his lab’s data on ApoE and effects on amyloid deposition prior to plaque formation (see ARF related news story). Mucke summarized his group’s positions as follows: ApoE3 prevents and ApoE4 promotes cognitive decline; E4 fragments play a significant role in this. ApoE4 plays a role in plaque-independent mechanisms. ApoE clearly also has an effect on plaque formation, but it is less clear that plaques and their associated neuritic dystrophy are responsible for cognitive decline.

Deposition or Clearance?
Taking a different view, David Holtzman of Washington University in St. Louis, Missouri, said that, to his mind, stopping or slowing the buildup of amyloid pathology remains the approach best supported by evidence. He cited longstanding observational studies of AD’s natural history at Washington University, led by John Morris and others. These studies indicate that people who die with the earliest hints of cognitive problems already have abundant amyloid pathology, indeed, almost as much as do people who die with moderate AD. Together with studies of Down’s patients, who develop plaques as teenagers but AD-like dementia around age 40, this suggests that amyloid deposition creeps up silently for up to 20 years before symptoms set in.

ApoE and its cousin ApoJ clearly play a central role in both amyloid deposition and clearance, but how that can be exploited therapeutically is less obvious, Holtzman said. Both these lipoproteins influence the still poorly understood transition from physiological, metabolized Aβ to the pathologic, insoluble variety that changes its conformation and aggregates in blood vessel walls and parenchymal spaces (see ARF related news story; see also ARF news story; scroll to Holtzman in ARF news story). ApoE’s triple roles in Aβ clearance, deposition, and blood/brain transport are not yet clearly enough delineated to suggest a strong therapeutic strategy, Holtzman said. Also still elusive is convincing in-vivo evidence of what is the most damaging Aβ species.

The mouse data are complex. In APP-transgenic mice lacking ApoE, Aβ deposits are diffuse, not fibrillar, and neither neuritic plaques nor dystrophic neurites develop. This would suggest that decreasing ApoE levels could help slow the pathology. On the other hand, PD-APP mice lacking both ApoE and clusterin (i.e., ApoJ), in which one cannot compensate for the other, deposit Aβ early and massively. This suggests that ApoE and ApoJ exert a key effect through clearance, and that increasing their levels might do some good. A complicating factor is that mouse ApoE and human ApoE appear to affect Aβ deposition/clearance differently, Holtzman added. Put simply, mouse ApoE tends to favor Aβ deposition, while all human ApoE isoforms, as well as the ApoE/ApoJ combination, tend to favor its clearance, E4 being the least efficient at that. Inducible promoter systems and gene therapy approaches should be used in mice to sort out whether increasing or decreasing ApoE and ApoJ levels is the way to go with drug development, Holtzman said. Another opportunity lies in drugs that affect the lipoprotein-Aβ interaction.

Reviewing results of a recent study of ApoE effects on Aβ deposition after a brain injury (see ARF related news story), Holtzman added that his lab is currently extending this work by studying graded, subtler injuries.

A major challenge ahead is to understand Aβ metabolism, Holtzman said. What happens after cells have secreted the peptide? How does ApoE and aging affect its distribution and concentrations in interstitial fluid, CSF, and blood?

Holtzman also urged labs studying ApoE to use physiological, CNS-derived ApoE preparations. He stated his view that a majority of available studies indicate that astrocytes are the major source of ApoE in the brain.

Ron DeMattos of Lilly Research Laboratories in Indianapolis presented a new study assessing the role human ApoE3 plays in the metabolism and deposition of Aβ in transgenic mice. Working in collaboration with Holtzman’s group at Washington University, and Steven Paul’s group at Lilly, DeMattos analyzed mice that overexpress human APP, have their own mouse ApoE knocked out, and express either no, one, or two copies of human ApoE3. The purpose was to measure whether the amount of ApoE3 mattered to the speed of Aβ clearance from the brain, and whether that translated into changes in how much Aβ the mice deposited in AD-relevant brain regions. To do that, the scientists measured soluble and insoluble Aβ levels in three- to six-month-old mice (which have not yet deposited Aβ) and in 12- to 15-month-old mice (which have). For both ages, DeMattos et al. measured Aβ levels separately in plasma and in the CSF; they extracted soluble and insoluble Aβ from the hippocampus and the cortex, and also stained fixed brain sections for diffuse and neuritic plaques. Gathering and analyzing this data took four years, but taking a simultaneous look at all these compartments is necessary to understand how ApoE3 dose affects Aβ metabolism, DeMattos said. The study showed that having higher E3 levels strongly reduced soluble Aβ levels before plaques appeared, and also shifted the ratio of CSF to plasma Aβ. Likewise, older mice with a double dose of ApoE3 had far fewer Aβ deposits, suggesting that ApoE3 protects against Aβ deposition by easing its traffic out of the brain (DeMattos et al., 2002, Soc Neurosci Abstract 32:723).

This begs the question of whether boosting ApoE levels might be a therapeutic strategy. Caveats here include how ApoE3/4 and 4/4 carriers would respond to such a treatment, DeMattos said. A similar study of Aβ pools with mice expressing different amounts of ApoE4 remains to be done. Prior studies hint that all ApoE isoforms are somewhat protective against Aβ deposition, but since ApoE4 also has toxic effects, increasing its expression is questionable. Another hitch lies in the question of when best to treat. For example, it is conceivable that a drug that increases ApoE levels could help before Aβ deposits have formed, but might accelerate further deposition if given later in the disease when Aβ levels in the brain already have become very high. Others warned that elevating ApoE3 beyond a certain level could accelerate atherosclerosis. Conditional transgenics that allow switching on ApoE3 expression at will could address the question of timing, DeMattos said.

Prior work already showed that CSF and plasma Aβ are connected in a dynamic equilibrium, and that the process of Aβ deposition blunts this gradient. The mechanisms of degradation and export of Aβ are complex and probably influenced by multiple Aβ-binding proteins, not just ApoE and ApoJ, DeMattos added. (See also see DeMattos et al, 2002; ARF related news story; ARF Live Discussion). On balance, DeMattos and Holtzman said they were not ready to venture an opinion on whether pharmacologically increasing or decreasing ApoE would make a better therapy.

Thomas Wisniewski of New York University School of Medicine, as well as others, noted that the Gladstone approach of finding small-molecule drugs that convert ApoE4’s structure to a more E3-like conformation are promising precisely because they are independent of ApoE’s complex A-b interactions.

I Spy: More Spines in E3, Fewer in E4
Wisniewski presented his lab’s experiments aimed at analyzing the observation that ApoE4/4 carriers, even those without AD, appear to suffer a more pronounced cognitive decline as they age than do ApoE3 or 2 carriers. This finding came out of neuropsychological tests and brain imaging, and Wisniewski is looking for a correlate at the level of individual dendrites. Using a gene gun to shoot fluorochromes into brain slices, his group worked out a protocol to label and observe synapses forming in the dentate gyrus, a target area of the perforant pathway that declines early on in AD. Two-photon microscopy then visualizes labeled dentate gyrus dendrites and individual spines on them. (Dendritic spines are highly fluid structures that respond quickly to activity and change during learning and memory; see ARF related news story and ARF news story.)

To study whether ApoE isotypes influence this process in aging mice, Wisniewski used Holtzman’s series of human ApoE transgenic mice. He took brain sections from these mice at three weeks, one year, and two years of age, labeled neurons with a fluorescent dye, and imaged their dendritic spines. Neither spine number nor spine density differed at three weeks, but both showed a decrease in ApoE4-transgenic or ApoE-knockout mice at one and at two years.

Wisniewski is also trying to replicate this finding in humans with an ongoing study of rapid (three-hour) postmortem tissue of AD cases, as well as age-matched controls who had neither cognitive problems nor pathology when they died. Among the AD group, ApoE4/4 carriers had the lowest spine density, and 3/3 carriers the highest. In this AD group, all brains had amyloid pathology, which could have caused the effect, as well. Wisniewski said he considers this confounder unlikely as even the normal aged brains showed the same difference in spine density between ApoE4 and E3 genotypes. He interprets these data as supporting the hypothesis of ApoE’s effect on synaptic reorganization and plasticity, which was originally formulated by Lissa Jarvik and Ashford in 1985, and later brought to light by Marsel Mesulam and Thomas Arendt.

ApoE in Hotbed of Neuroinflammation?
Linda Van Eldik of Northwestern University in Chicago talked about the role of ApoE in neuroinflammation. Prior work in her lab has examined the antiinflammatory effect of ApoE in suppressing Aβ’s ability to activate glial cells. However, more recent work in collaboration with MaryJo LaDu, ENH Research Institute in Illinois, is complicating the picture by suggesting that ApoE can be either pro- or antiinflammatory.

Abundant evidence indicates that inflammation fuels the progression of AD pathology, Van Eldik said. Activated microglia and astrocytes surround and infiltrate plaques, where they release inflammatory mediators such as interleukin-1b (IL-1b) and inducible nitric oxide synthase (INOS), among others. Polymorphisms in cytokine genes are beginning to be associated with disease susceptibility, and antiinflammatory treatment appears protective in epidemiological studies and in mouse models (though not so far in human treatment trials). Van Eldik’s lab works to identify the signaling pathways underlying this inflammatory component, and how ApoE becomes induced and then changes these pathways; the goal is to find ways to modulate those pathways with new antineuroinflammatory drugs.

Van Eldik proposes a model whereby glial proinflammatory factors cause neuronal dysfunction, which signals further glial activation and revs up a cycle of excessive glial activation and neuronal damage. Aβ accelerates this, as it damages neurons but also gets taken up into both microglia and astrocytes. Van Eldik’s group reported years ago that, in vitro, ApoE3 and 4 interfere with this cycle by blocking glial activation in response to Aβ (see Hu et al., 1998). ApoE3 and 4 are somewhat selective; they do not block the glial response to other activators such as IL-1b or cAMP. With LaDu and others, Van Eldik confirmed ApoE’s antiinflammatory effect in glia cultured from knockout mice. She suggested that ApoE receptors on glia do two opposing things: LRP mediates Aβ-induced glial activation, and the LDL-receptor mediates Aβ-induced increases in ApoE levels, which in turn dampen the inflammatory stimulation inside the glial cell.

Up to this point, there were no clear isoform effects. That changed, however, with more recent experiments, which have added a new wrinkle by indicating that ApoE can also be proinflammatory in an indirect way. The Aβ-induced ApoE expression in glial cells mentioned above is blocked by exogenous ApoE added to the cultures, and ApoE4 does this more effectively than does E3. And when there is no Aβ present, ApoE4 (much more than ApoE3) stimulates IL-1b production.

In summary, this in-vitro work suggests that excess Aβ, signaling through ApoE receptors on glia, stimulates these cells, and at the same time induces ApoE expression, which serves to control their activation. Yet overproduction of ApoE in highly activated glia can exacerbate proinflammatory pathways.

With regard to developing ApoE-based therapeutics for AD, these observations imply that future inhibitors of ApoE production or activity may be a useful strategy for people who carry the ApoE4 allele. Specific small-molecule inhibitors that could suppress ApoE’s proinflammatory activity without affecting its antiinflammatory activity would be extremely attractive. For those who carry E2 and E3 alleles, development of an ApoE agonist or mimetic to boost the neuroprotective effects of these proteins might also be feasible. Whatever the eventual approach, more research on the molecular mechanisms by which ApoE modulates the neuroinflammatory and neuropathologic cascade is critically needed in the field as a foundation for future development of any ApoE-based drugs against AD, Van Eldik said. This research must include elucidation of the glial signaling pathways that ApoE4 vs. E3 can induce, delineation of the molecular mechanisms by which ApoE modulates ApoE3 and 4-induced glial and neuronal responses, and exploration of points of intervention in the pathways to identify new drug discovery targets for preventing or treating AD.

MaryJo LaDu extended these findings by summarizing studies of the structural and functional interactions between ApoE and Aβ. She first described experiments to determine which forms of Aβ cause the glial activation that ApoE appears to modulate. The question of whether experimental preparations of particular Aβ forms are stable or convert from one form to another and form mixtures is frequently debated. LaDu’s lab has made stable preparations of monomer, oligomer, protofibrils, and fibrils (see Stine et al., 2003), and uses those to study neuroinflammatory interactions between Aβ and ApoE. LaDu showed in-vitro data suggesting that oligomers are more potent than fibers at inducing neurotoxicity, and at inducing the morphological changes, nitric oxide, and INOS induction that mark glial activation.

LaDu also presented electrophysiologic data gathered in collaboration with Barbara Trommer. In hippocampal slice cultures, ApoE3 prevented oligomer-induced neurotoxicity and impairment of long-term potentiation in rats, whereas ApoE4 did not. Injecting Aβ oligomers into ApoE-knockout mice disrupted LTP; adding back ApoE3, but not E4, restored LTP. Similar experiments comparing LTP in ApoE3- and E4-transgenic mice, knockout, and wild-type mice, indicate that ApoE3 has a protective effect but ApoE4 does not.

In summary, LaDu said she suspects that ApoE interacts with Aβ oligomers to speed their clearance and thus inhibit neurotoxicity and inflammation, and that E3 does so more effectively than does E4. She agrees with Holtzman and others that ApoE facilitates amyloid deposition, E4 more so than E3. Future small-molecule drugs could mimic ApoE3 binding, she said. In this context, LaDu mentioned a monoclonal antibody, 7A2, that is specific to Aβ oligomers. Her lab is currently studying this interaction in detail to generate information that could guide drug development. As of now, however, LaDu cannot yet say which parts of Aβ oligomers bind to ApoE, as this varies with Aβ’s aggregation state and ApoE’s lipidation state. Both proteins have extensive tertiary structure, and better-defined conditions are needed to study their interaction.

The Insulin Connection
Taking a different tack, Steven Edland of the Mayo Clinic in Rochester, Minnesota, presented case-control data that associate ApoE with insulin-degrading enzyme (IDE) in Alzheimer’s. IDE is among a number of proteases that can degrade Aβ. It has become a focus in a broader evolving story implicating insulin signaling pathways in aging and neurodegeneration. Two recent mouse studies have shown an increase in brain Aβ in IDE-knockout mice (see ARF related news story), and on the human front, the IDE gene is still in the running for being the elusive risk factor on chromosome 10q (see ARF related news story). However, two association studies, one led by Leslie Jones (see ARF related news story) and one by Boussaha et al. 2002), found no association between IDE variants and AD. Edland decided to address the question partly because Jones’s study had not stratified by ApoE genotype, and partly because other links among insulin resistance, ApoE genotype, and AD kept emerging in the literature. The story is confusing, however, as the literature does not all fit into a coherent picture. Some studies connect hyperinsulinemia to ApoE3 AD patients (see Kuusisto et al, 1997 and Craft et al., 1999), while others find an association between diabetes (a frequent consequence of insulin resistance) and ApoE4 AD patients (see Peila et al., 2002). Another recent study found lower IDE expression in hippocampus of ApoE4-carrying AD patients than that of other genotypes (see Craft section in ARF related news story).

Edland and colleagues investigated this question in a sample of 80 normal controls and 118 cases recruited by the Mayo Clinic AD patient registry. He found three SNPs (IDE1, IDE3, IDE8) that were associated with AD among patients with ApoE3, while none of the studied SNPs associated with AD among patients who carried ApoE4. Haplotype analysis indicated that the protective alleles fell into a block that drove this association. The IDE1 and IDE3 variants are in the noncoding promoter region of ApoE, and therefore are more likely to affect expression levels than protein activity, Edland said. IDE8 resides in an intron and could affect mRNA splicing.

Edland cautioned that his result-like all results from genetic association studies-could be a false-positive, and needs validation before it can be interpreted. However, if lower hippocampal IDE expression is relevant to the development of AD, and if that is driven by ApoE4 in people with the E4 allele (see Craft section in ARF related news story), then other causes of lower IDE expression or activity, such as IDE genetic variants, could plausibly explain increased risk in subjects without an E4 allele. The effect of ApoE4 on IDE expression might be mediated by cortisol, which is elevated in ApoE4 carriers (Peskind et al., 2001). Chronic glucocorticoid administration lowers hippocampal IDE expression levels in the macaque monkey, Edland noted, suggesting that treating the hormonal stress axis might be of benefit. Mucke noted that it is difficult to separate a specific ApoE-IDE effect from general anxiety, which is known to afflict many AD patients and even some mouse models of b-amyloidosis. It could also be that the genetic IDE association is real, yet does not act through ApoE; rather, certain ApoE isoforms could improve a person’s ability to withstand stress.

Therapeutic implications of this IDE work include treating AD with PPARg agonists, such as the thiazolidinedione drugs pioglitazone and rosiglitazone, which doctors are routinely prescribing for insulin-resistant diabetes(see ARF related news story). Indeed, Allen Roses mentioned that GlaxoSmithKline is currently conducting a follow-up phase 2 trial of rosiglitazone in Europe. This drug was picked because it does not readily enter the CNS. PPARg agonists support brain glucose uptake, and have antiinflammatory effects, (see Alzforum Live Discussion). An abstract by GSK scientists last November in Orlando reported that PPARg agonists block Aβ-induced expression of proinflammatory cytokines by microglia and monocytes (Richardson et al., 2002), Roses also summarized expression proteomics studies comparing ApoE-knockout and wild-type mice. The proteins in which levels were either increased or decreased fell into broad categories that indicate areas for therapy development irrespective of underlying mechanisms or hypotheses, he said. Surprisingly many of the changed proteins had to do with glucose metabolism and bioenergetics, Roses said, and that is where the company is now focusing its development efforts.

First Things First: Don’t Forget Cholesterol
Bill Rebeck of Georgetown University in Washington, DC, introduced CYP46 and LXR as potential drug targets that lie upstream of ApoE and relate to cholesterol export out of neurons. Rebeck’s approach is inspired by the observation, up and coming in recent years, that elevated cholesterol increases one’s risk for AD and Aβ deposition. One approach to studying molecular consequences of elevated cholesterol focuses on its effect on membranes and APP processing (Wolozin, 2001; Puglielli et al., 2003). Rebeck chose instead to focus on the gene expression the cell uses in adjusting its cholesterol levels. He borrows information learned in cardiovascular research and looks for similarities and differences in the brain. The brain tries to reduce excess intracellular cholesterol by first converting it to 24S-hydroxycholesterol. The CYP46 member of the cytochrome P450 family of enzymes performs this step, and has, indeed, been implicated as an AD risk factor itself, though other studies, including Rebeck’s, have not confirmed this finding (see ARF related news story). This 24S-OH form of cholesterol is normally found only inside the brain, and its CSF levels are considered a possible early marker of incipient AD and also of statin treatment (see ARF related news story).

24S-hydroxycholesterol forms a complex with the nuclear hormone receptors LXRb (the version of LXR in the brain) and RXR. This complex then acts as a transcription factor and increases expression of genes that, in turn, reduce cellular cholesterol levels. One of these induced genes is ApoE. Rebeck reported that LXR agonists upregulate ApoE in cultured microglia, but said that LXR is most likely responsible for ApoE induction in special circumstances, not for its basal expression. Part of the explanation for the elevated AD risk in ApoE4 carriers may be that E4 binds cholesterol less readily than E3, and generally is less efficient at getting it out of the cell, Rebeck said.

Another cholesterol-efflux gene induced by the LXR/RXR complex is ABCA1, Rebeck said. To study whether LXR influences Aβ production, Rebeck’s group treated cultured cells with a synthetic LXR agonist and found that it increased not only ABCA1 and ApoE in glia but also the amount of Aβ42 the cells secreted (Fukumoto et al., 2002). How that LXR-Aβ42 connection comes about is unclear.

From these data, Rebeck suggested the design of therapeutic CYP46 or LXR antagonists to reduce Aβ generation. Such compounds could, however, also increase intracellular cholesterol, as they would interfere with the cell’s cholesterol efflux mechanism. This rationale starts from the assumption that elevated cholesterol levels-either from the diet or from ApoE4/4 genotype-is the original insult that throws a finely calibrated system out of whack, Rebeck said.

Not surprisingly, at the end of this intense day, opinions differed about where to go from here. Some noted that there was still no cohesive, overarching story to integrate the myriad, sometimes conflicting hypotheses for ApoE’s role in AD, and that this made it difficult to decide where to interfere therapeutically. Others felt that the diversity of mechanisms need not be a problem. ApoE likely functions in many different pathways, each of which has meaning in the complex beast that is brain metabolism. While, indeed, it is not clear yet which ones are of overriding concern in the early disease state, candidate pathways for therapeutic exploration do exist, these scientists said. Mahley’s and Weisgraber’s domain interaction busters appeared to draw majority support for ApoE-based drug leads. Finally, the field appears to begin generating overlapping results-for example, E3 protects better than E4, plus E4 is bad on its own; or ApoE is at the nexus of clearance and deposition-that will help the picture to gel. One thing everyone agreed on is that, considering ApoE’s unrivaled genetic influence on late-onset AD, more laboratories still need to join the fray.—Gabrielle Strobel

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

News Citations

Paper Citations

Corder EH, Saunders AM, Strittmatter WJ, Schmechel DE, Gaskell PC, Small GW, Roses AD, Haines JL, Pericak-Vance MA. 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. PubMed.
Silverman JM, Smith CJ, Marin DB, Mohs RC, Propper CB. Familial patterns of risk in very late-onset Alzheimer disease. Arch Gen Psychiatry. 2003 Feb;60(2):190-7. PubMed.
Corbo RM, Scacchi R. Apolipoprotein E (APOE) allele distribution in the world. Is APOE*4 a 'thrifty' allele?. Ann Hum Genet. 1999 Jul;63(Pt 4):301-10. PubMed.
Teter B, Ashford JW. Neuroplasticity in Alzheimer's disease. J Neurosci Res. 2002 Nov 1;70(3):402-37. PubMed.
Huang Y, Liu XQ, Wyss-Coray T, Brecht WJ, Sanan DA, Mahley RW. Apolipoprotein E fragments present in Alzheimer's disease brains induce neurofibrillary tangle-like intracellular inclusions in neurons. Proc Natl Acad Sci U S A. 2001 Jul 17;98(15):8838-43. PubMed.
Xu PT, Gilbert JR, Qiu HL, Ervin J, Rothrock-Christian TR, Hulette C, Schmechel DE. Specific regional transcription of apolipoprotein E in human brain neurons. Am J Pathol. 1999 Feb;154(2):601-11. PubMed.
Raber J, Bongers G, LeFevour A, Buttini M, Mucke L. Androgens protect against apolipoprotein E4-induced cognitive deficits. J Neurosci. 2002 Jun 15;22(12):5204-9. PubMed.
Demattos RB, Bales KR, Parsadanian M, O'Dell MA, Foss EM, Paul SM, Holtzman DM. Plaque-associated disruption of CSF and plasma amyloid-beta (Abeta) equilibrium in a mouse model of Alzheimer's disease. J Neurochem. 2002 Apr;81(2):229-36. PubMed.
Hu J, LaDu MJ, Van Eldik LJ. Apolipoprotein E attenuates beta-amyloid-induced astrocyte activation. J Neurochem. 1998 Oct;71(4):1626-34. PubMed.
Boussaha M, Hannequin D, Verpillat P, Brice A, Frebourg T, Campion D. Polymorphisms of insulin degrading enzyme gene are not associated with Alzheimer's disease. Neurosci Lett. 2002 Aug 23;329(1):121-3. PubMed.
Kuusisto J, Koivisto K, Mykkänen L, Helkala EL, Vanhanen M, Hänninen T, Kervinen K, Kesäniemi YA, Riekkinen PJ, Laakso M. Association between features of the insulin resistance syndrome and Alzheimer's disease independently of apolipoprotein E4 phenotype: cross sectional population based study. BMJ. 1997 Oct 25;315(7115):1045-9. PubMed.
Craft S, Asthana S, Schellenberg G, Cherrier M, Baker LD, Newcomer J, Plymate S, Latendresse S, Petrova A, Raskind M, Peskind E, Lofgreen C, Grimwood K. Insulin metabolism in Alzheimer's disease differs according to apolipoprotein E genotype and gender. Neuroendocrinology. 1999 Aug;70(2):146-52. PubMed.
Peila R, Rodriguez BL, Launer LJ, . Type 2 diabetes, APOE gene, and the risk for dementia and related pathologies: The Honolulu-Asia Aging Study. Diabetes. 2002 Apr;51(4):1256-62. PubMed.
Peskind ER, Wilkinson CW, Petrie EC, Schellenberg GD, Raskind MA. Increased CSF cortisol in AD is a function of APOE genotype. Neurology. 2001 Apr 24;56(8):1094-8. PubMed.
Wolozin B. A fluid connection: cholesterol and Abeta. Proc Natl Acad Sci U S A. 2001 May 8;98(10):5371-3. PubMed.
Puglielli L, Tanzi RE, Kovacs DM. Alzheimer's disease: the cholesterol connection. Nat Neurosci. 2003 Apr;6(4):345-51. PubMed.
Fukumoto H, Deng A, Irizarry MC, Fitzgerald ML, Rebeck GW. Induction of the cholesterol transporter ABCA1 in central nervous system cells by liver X receptor agonists increases secreted Abeta levels. J Biol Chem. 2002 Dec 13;277(50):48508-13. PubMed.

Institute for the Study of Aging Symposium

Tools

PET Diagnosis Poised for Prime Time? FDA Wants Consensus, Better Trials

Comments

Make a Comment

References

News Citations

Paper Citations

External Citations

Further Reading

Papers

News

ApoE Catalyst Conference Explores Drug Development Opportunities

Comments

Make a Comment

References

News Citations

Paper Citations

Other Citations

External Citations

Further Reading