Chapter 5

Alzheimer Genes: Biomarkers of Prediction and Prevention

Every morning she wakes up a little further away from us. Yet she still manages to convey something, if only with a glance, of the world she is entering. Beyond the fear and the loss, she seems to say, there is life of a sort here, at the dark edge where everything is crumbling.

—Michael Ignatieff, Scar Tissue¹

In a 1997 article titled “Plundered Memories,” Zaven Khachaturian, then the director of the Ronald and Nancy Reagan Research Institute of the Alzheimer’s Association, and later director of the Office of Alzheimer Disease Research Center at NIH, wrote,

Some critics ask whether genetic research is worth the resources it consumes and the anguish it will bring to those who test positive for a harmful gene—when a cure still seems so far away. In my view, however, the genetic approach is on the right track, and I think the continuing research on Alzheimer disease may soon confirm that belief. Those of us in the front lines of the fight against Alzheimer’s have never been closer to unmasking this mysterious thief, the robber of the very thing that makes human beings unique.²

In this article, Khachaturian is scornful of the evolutionary biologist Richard Lewontin who had recently suggested that genetic research into Alzheimer’s is like going down a blind alley as far as cures are concerned (a comment that to date has proven to be the case). Khachaturian, passionate about the need to combat Alzheimer disease, has devoted much of his life to this cause, and raised millions of dollars for research. His mother died of probable AD, and over a decade ago Khachaturian started to think that he himself was beginning to suffer from it; in retrospect it seems that this is not so.

Given the findings about AD genetics that had appeared just prior to the publication of “Plundered Memories,” the apparently extravagant claims made by Khachaturian do not seem so wildly out of line, Lewontin’s sagacious comments notwithstanding. It is of note that the early articles about the discovery of genes “causal” of AD raised many of the questions and uncertainties that continue to plague the AD world today. Moreover, their discovery precipitated the formulation of the amyloid cascade hypothesis that continues to be the dominant model of AD causation.

Dominantly Inherited Alzheimer’s

In a 1996 article, the neurologists Ephrat Levy-Lahad and Thomas Bird pointed out that the high prevalence and late age of onset of the common form of Alzheimer disease are features not usually associated with “genetic disease.” This was an acceptable statement in the days before molecular genetics had become firmly entrenched, but the authors then went on to self-consciously rebut their own statement, presaging some of the dramatic changes about to take place in the world of genetics as a whole. The body of their article was a discussion of three autosomal dominant genes³ that, using linkage studies followed by positional cloning had, in the 1990s, been shown to be causal of what is today termed “familial Alzheimer disease” or “dominantly inherited Alzheimer’s disease.” The first of these genetic loci to be uncovered, the amyloid precursor protein (APP) encodes the β-amyloid peptide associated with the buildup of amyloid plaque (see chapter 2). Originally four mutations of this gene were found, but more than 30 are now recognized, at least 25 of which cause dominantly inherited AD. Individuals are usually affected between ages 40 and 60.⁴ It was relatively recently recognized that the APP gene has what may well be a critical role in learning and memory function during development, and that maintaining an appropriate level of APP expression is important but, if disrupted, then excess β-amyloid may result, leading to plaque formation.⁵ It is currently estimated that APP mutations cause approximately 15% of all cases of familial AD.

Two related genes known as presenilin-1 and presenilin-2 are also associated with dominantly inherited Alzheimer disease. Presenilin-1, first mapped fifteen years ago, is involved with brain and spinal cord development and with the processing of APP into smaller fragments. Over 185 mutations of presenilin-1 have been shown to affect more than 30 large extended families worldwide, with a very early age of onset, on average between 35 and 55 years. These mutations are associated with plaque and tangle pathology, and account for approximately 70% to 75% of the cases of familiar AD.

The presenilin-2 gene was discovered when research was carried out among Volga German families living in the United States—families who emigrated from Germany to Russia in the 1760s, many of whom eventually migrated to the United States. A second large extended family that carries this mutation is Italian.⁶ Presenilin-2 is involved with cell growth and maturation as well as the processing of APP into peptide fragments. Presenilin-2 is also associated with plaque and tangle pathology. At present, 13 known mutations are associated with this gene, accounting for approximately 5% to 10% of dominantly inherited Alzheimer cases.

Presenilin-2 mutations are not strictly speaking autosomal dominant because a few individuals who carry these genes do not develop AD, even at a great age. Furthermore, although the mean age of onset of presenilin-2-initiated AD approaches 55 years, it is striking that the age range can vary from 40 to 75 years, even among families who carry identical mutations. For familial AD in general it has been shown that the age of onset for identical twins can vary by as much as a decade, strongly suggesting that environmental factors influence to some degree the “penetrance” (phenotypic expression) of these genes.⁷ Research indicates that these 230 or so presently known mutations associated with the APP gene, presenilin-1, and presenilin-2 account for between 5% and 10% of all AD cases (this figure is a rough estimate given that, globally, cases of Alzheimer’s, even cases of early onset Alzheimer’s, are not inevitably diagnosed). Yet more mutations are regularly confirmed.

Using new sequencing technology, it is now apparent that rare mutations of all three of the genes associated with dominantly inherited AD are at times present in individuals not diagnosed with dementia until later in life. This research was carried out with individuals who come from families where four or more members are affected with AD. This finding strongly suggests that an unknown factor or factors must function to defer the age of onset of dementia in these instances.⁸ Increasingly, as research into AD genetics advances, it is clear that boundary making among what are thought to be different forms of AD on the basis of age of onset, as has usually been the case, may not be highly informative.

It is usually stated that it was a dominantly inherited form of AD that caused dementia in Alois Alzheimer’s patient Auguste D, but there is no conclusive evidence that this was so. It is known that a good number of cases of so-called sporadic or late-onset AD can affect individuals at a relatively young age, but it will not be possible to establish definitively what “type” of AD so tragically ended Auguste D’s life.

In an editorial published in 1996 written for the Annals of Neurology, Bradley Hyman pointed out that during the course of the 1980s research had shown that the neuropathological features—the plaques and tangles—of early-onset dominantly inherited AD (known as “presenile dementia” throughout most of the last century) were essentially identical to those in the more common form of late-onset Alzheimer’s. With the discovery of the genes associated with familial Alzheimer disease, Hyman suggested that a reevaluation was required as to whether or not early-onset AD and late-onset AD are indeed a single process happening on different time scales or, alternatively, “do these various genetic factors produce a group of ADs that share some phenotypical commonality but are distinct processes?”⁹ As we will see in later chapters, this question continues to be of great significance, particularly in connection with drug development.

Hyman recognizes that his concern is, in effect, a “philosophical question”—a matter of ontology—but he uses findings from molecular epidemiology to conclude that genetic factors associated with familial AD, while they may result in what appears to be a worse disease, do not apparently culminate in a qualitatively different pattern of brain involvement. In other words, β-amyloid deposition seems to be the neuropathological factor most strongly influenced by the known genetic factors, and ultimately this leads to “ ‘Alzheimer disease’ rather than ‘Alzheimer diseases.’ ” Hyman concludes, “In this way, Alzheimer’s is probably similar to atherosclerosis and other complex trait diseases in that there are multiple possible influences, some genetic, some ‘familial’ and only defined in a vague way, and potentially, even some environmental. These all, presumably feed into a common pathophysiological process that leads to a very similar clinical and neuropathological phenotype.”¹⁰

John Hardy and Gerald Higgins published their hypothesis about the amyloid cascade in Science in 1992, and it was the discovery of the APP gene, details about which were published in Nature in 1991, that made their hypothesis so appealing to numerous researchers. Discovery of the genes associated with dominantly inherited Alzheimer disease also ensured that those families in which the genes were present would be regarded as desirable research subjects, particularly in connection with drug development designed to stop amyloid deposition. Interest in recruiting members of these families as research subjects has come to a head in the past few years, particularly with the move to bring about AD prevention.

The Paisa Mutation

Little by little, studying the infinite possibilities of a loss of memory, he realized that the day might come when things would be recognized by their inscriptions but that no one would remember their use.

—Gabriel García Márquez, One Hundred Years of Solitude¹¹

A large group of 25 extended families, originally of Basque origin, about 5,000 in all, live in both urban and rural areas of the province of Antioguia, Colombia, scattered mostly around Medellín. These families compose the most concentrated group of individuals known to date who carry a specific mutation of the gene presenilin-1. Close to one-third of the population—approximately 1,500 people—harbor what is known locally as paisa—a word that simply refers to the people of the region. Typically, memory loss becomes manifest around 45 years of age, and most affected individuals progress to full-blown dementia by their early 50s—a very rapid progression to La Bobera (the foolishness). Francisco Lopera, a neurologist associated with Antioquia University in Medellín, was the first scientist to pay serious attention to the devastating condition that affected so many Antioguian families. Once Lopera understood that during the course of three generations, half of the children were affected, he realized that he had stumbled across a form of dementia that must surely be due to a single gene mutation. Only a decade later, in the 1990s, was the gene isolated, at which time the finding began to attract international attention. Despite the serious risks posed by drug traffickers and Farc rebels, who were very active in the 1980s and ’90s, Dr. Lopera and his colleagues managed to collect DNA samples and assemble a genetic pedigree extending back nearly 300 years, to around the time that these families first arrived as part of the ongoing colonization of Colombia by Spaniards, searching for gold.

In late 2010, the executive director of the Banner Alzheimer’s Institute in Phoenix, Arizona, Eric Reiman, a newly recognized rock star of science, appeared on Fox News following an appearance in GQ magazine together with the reality star Bret Michaels in a Los Angeles studio. This event took place in order to raise money for Alzheimer’s research, some of which would be directed toward a planned trial with the Colombian families in collaboration with Francisco Lopera in Medellín. A 2011 BBC news commentary noted that one scientist (unnamed) described these extended families as a “natural laboratory.”¹² Researchers are excited, listeners were told, because the interrelated families provide them with an “enriched sample.” Earlier, Neil Buckholtz of the NIA informed a reporter from the New York Times that these individuals will have “cleaner brains that can give a better picture” than a sample made up of older subjects.¹³ In other words, because an autosomal dominant mutation is involved, it is virtually certain that at least one-third of the subject population will become demented and, significantly from the point of view of the researchers, at a young age.

This same newscast reported that the Banner Alzheimer’s Institute planned to use subjects selected from among the Colombian families to test drugs designed to attack plaque in the brain as part of a newly formulated Alzheimer’s Prevention Initiative (API). It is acknowledged in the newscast that it is not known if amyloid plaque is the cause or an effect of Alzheimer’s, but, as Joseph Arboleda, a Harvard-based researcher interviewed by the BBC who works with Dr. Lopera, stated, “the trial puts this hypothesis to the test. … It is possible that drugs will inhibit the brain plaque and yet the family will still get dementia. Such results would prove devastating for current research.” However, “If in the extended family the onset of Alzheimer’s is delayed, or stopped, then the researchers will have hit the mother lode—a potential cure for sufferers worldwide. That remains a big if.”¹⁴

Eric Reiman has also been very explicit in suggesting that, should “the disease be halted,” this “could generate treatments to protect millions worldwide from common Alzheimer’s.”¹⁵ Commenting on the proposed trial in 2011 Reiman said,

[I]f there is no effect on the biomarkers after 2 years in the right direction [we will] declare futility and give these people at the highest imminent risk access to the next promising treatment. If, however, they do budge in the right direction [we will] continue to follow them a little bit longer to see if it slows even subtle memory decline … if it does, I have a feeling that may be enough evidence for regulatory agencies to consider using biomarkers under their accelerated approval mechanism in other [AD] populations.¹⁶

Reiman does not make clear what would be the “next promising treatment,” should the trial drug prove ineffective. In 2012 a Reuters report was emphatic that “[t]he trial in Colombia could offer the most definitive test yet of the amyloid theory of Alzheimer’s. … This trial [will] be different because it will be tested on people before the disease has done much damage to the brain cells.”¹⁷ The excitement associated with this research is remarkable, and I was told informally in the summer of 2012 that Francisco Lopera was receiving so many email communications from interested parties of one kind and another that the Banner Alzheimer’s Institute scientists had taken it upon themselves to protect him as best they could from this onslaught.

Very moving stories accompanied by striking images about elderly parents caring for their affected adult children accompanied both the BBC news program and the New York Times report. These accounts are disturbing, not only because of the pathos involved, but also because the trial raises many troubling ethical questions. Eric Reiman is reported to have said that his team is “trying to manage expectations within the Colombian families, making clear the drug is experimental and risks and benefits are uncertain,”¹⁸ something he reiterated with me on the phone. When I spoke in July 2012 with Pierre Tariot, an experienced geriatric neurologist and an associate director of the Banner Alzheimer’s Institute, also involved with the trial, he told me that discussions have been taking place since 2006 about whether or not the trial was really feasible. Matters such as how best to “operationalize” this complex, potentially exploitative project were of great concern to involved researchers from the outset. Among the many questions posed were these: How can the trial be carried out in a way that would be “respectful and humble”? Should the biological materials obtained from these peoples be regarded as, in effect, a “sacred resource”?

A registry was set up in Medellín in 2012 enabling potential participants to make a connection with Dr. Lopera, and signatories were asked to attend seminars where the ultimate objectives of the trial; what would be required of research subjects, and the process of randomization were explained at length. Many of the trial subjects have at least 10 years of formal education, but others have only three years or so of schooling. Dr. Tariot noted that it was readily apparent that the information was attentively received and mutually discussed. Not everyone has proved eager to participate in the trial, but the community as a whole has accepted the plan very favorably, and many have expressed willingness to participate, even among those who have no signs of dementia.

These families are practicing Catholics, and, furthermore, young women are raised to believe that they should have children—that this is part of God’s will. Given this situation, it was important to involve local priests in discussions from the outset. One requirement of the study is that women should not become pregnant during the trial, and therefore should be using contraceptives. The priests proved to be cooperative, and apparently had no difficulty with contraceptive use for this purpose. The media reported that at least one woman of reproductive age had obtained a hysterectomy in order to avoid passing the disease on to the next generation—confirmation of the inescapable fears that people in this community confront, even when childbearing is considered so important.¹⁹ Kenneth Kosik, a neurologist from UC Santa Barbara who has been involved with the community since the 1990s, encountered a young woman who very much wanted to have children, but was terrified at the prospect of passing down the mutated gene.²⁰ Kosik further noted, when commenting on the possibility of genetic testing among this population, “It’s very dangerous knowledge … I saw a 23-year-old man who said that if he found out he had the mutation, he would commit suicide.”

In late 2011, members of a few of the affected Medellín-area families were brought to Phoenix where they were given PET scans, among other preliminary tests, that will provide a baseline for amyloid deposition in their brains before the trial commences. As of mid-2013 ethics clearance of the trial had not been completed; the intent had been agreed upon in writing by the FDA, but the trial still awaited clearance from both the Colombian Ministry of Health and the FDA. Should this trial go forward, it will be a novel approach to testing drugs for AD; prior to this time subjects have been identified as having dementia before entering trials.

Trial Preparation

The randomized trial, with an estimated cost of $100 million, will be coordinated out of the Banner Alzheimer’s Institute and is now slated to begin in late 2013 with the hope of producing interim results by 2017. Collection of data will be carried out entirely in Colombia, and a cyclotron has been set up for the first time at great expense in Medellín in order that neuroimaging can be carried out on-site. In 2012, the NIH announced that it will provide $16 million toward the cost—part of the U.S. government initiative taken by President Obama to provide funds for AD research. Another $15 million comes from private sources, and the drug company Genentech, the U.S.-based biotechnology unit of Roche based in Switzerland, will provide the biggest amount—$65 million. Selected study participants must be at least 30 years of age, and within 15 years of the age when their parents’ symptoms began. A total of 100 healthy individuals who carry the paisa mutation will be enrolled in the trial and be administered the drug Crenezumab, an experimental antibody that targets amyloid buildup (it is claimed that this drug has fewer side effects than other anti-amyloids used to date, and was selected from among “25 rivals”).²¹ The researchers point out that because the drug is less toxic, it can be used at higher doses, and will attack pathological forms of both soluble and insoluble forms of amyloid in the brain. The drug will be administered subcutaneously every two weeks.

Among the 200 controls, half will be individuals who carry the mutation and the other half who do not. All trial participants will be administered lumbar punctures, brain scans, and other tests at regular intervals over the course of 24 to 60 months to track biomarkers and other possible clinical changes unless, of course, the trial is stopped prematurely. Among the tests, several are designed to detect subtle signs of memory loss—described as the “primary outcome measure” of the trial. Noninvolved neurologists have made passing comments to me about the duress that these subjects will be under by submitting themselves repeatedly to these procedures, all the while receiving little or no communication about data findings. The expressed hope is to get an early indication that the drug is indeed removing amyloid. Participants will learn ahead of time that they will not be told whether or not they are carrying the mutation, even after the trial is completed. And it appears that the research subjects have no wish to learn this information, a position that causes the researchers to feel at ease, because no genetic counselors are available in Medellín. The three-armed trial will mean that the research subjects are “blinded,” that is, they will be unable to tell who among them carries the mutation.

It remains an open question as to what extent the research subjects fully comprehend the uncertainties associated with the trial in both its implementation and its outcome, despite the best efforts of the research team to explain matters. Given the tragedy that stalks the lives of all these families, it surely must be the case that many of these individuals would be willing to undergo extensive personal discomfort and risk for even a small possibility that a medication may be discovered that will lift the monstrous burden that their families must bear.

Pierre Tariot is convinced that this research will not result in harm and that, whatever the outcome, it will be informative. When he talked with me, he emphasized that millions of dollars had already been spent in preparatory work. Tariot noted that if benefit clearly results, then, of course, that will be the best possible outcome, but this may well not be the case. He is convinced that the dose level that will be administered of the selected drug is correct but that quite possibly it should be started at an even earlier age than 30 to be effective. He emphasized, in contrast to much media reporting, that should the drug prove effective, it could be used only for kindred of the Colombian families with the same mutation. Paisa is a founder mutation, one that was present among the families who made up the original Basque immigrants, or else arose shortly after their arrival in Colombia, thus substantiating Tariot’s opinion, that much more research would be required before it was known whether or not the particular drug being used in the trial could be applied to other populations. Eric Reiman, in common with many other AD experts, has made it clear to the media, in contrast to his colleague, that in his opinion dominantly inherited, early-onset AD is essentially the same condition as is late-onset AD in its final stages. Hence any drug that proves effective in stopping the former condition is likely to work in a similar manner as the latter. However, the position taken by John Hardy, the originator of the amyloid cascade hypothesis (see chapter 2), and increasingly other commentators, is that this may well not be the case. Preliminary findings from exploratory research with Colombia subjects suggest that the latter position may well be correct, as we will see shortly.

A Question of Beneficial Distribution

Concerns that will arise in the minds of social scientists in connection with this trial are reminiscent of problems associated with the Human Genome Diversity Project, and also with the expanding universe of drug trials in which “naïve” subjects (individuals who do not routinely imbibe medication) often living in economically deprived conditions are systematically recruited for trials. In recent years the “importation of subjects” to carry out research in “developed countries” has started to take place.²² This method is usually adopted when small, novel research strategies are implicated, such as gene transfer, or isolated populations of interest are involved, as is the case with the Medellín subjects for pretrial testing. After extended scrutiny in connection with both its science and the involved ethics, the Human Genome Diversity Project was, in the event, never funded. The strongest criticism of this project, had it taken place, and also of the exponential increase in the “offshoring” of clinical trials, is that the final “products” of these endeavors, if successful, will primarily benefit the developed world, and not the subjects under study.²³ The Colombian trial has been set up in such a way as to avoid this type of blatant exploitation. The arrangement with Genentech is such that, should the trial prove successful, medication will be made available to all affected families on an ongoing basis. Presumably too, local health care facilities will have to be expanded. The anthropologist Nikola Bagic, who has worked in the region, stresses that by far the majority of Colombians live in relative or extreme poverty and lack access to anything more than very basic primary health care. His ethnographic research among families who harbor the paisa mutation has made it clear to Bagic that their travails must first and foremost be recognized as a political matter. Pensions and other forms of social support do not exist, resulting in extraordinary stress for affected families, who inevitably become desperately impoverished particularly because the disease strikes young and middle-aged adults. However, Bagic is encouraged because the local physician/researchers based in Medellín have had considerable resources made available to them in preparation for the trial, resources that they plan to use not only to improve their own decrepit computers but also to improve family well-being (personal communication). It may well be that this potential influx of social support provides the greatest incentive of all for these families to participate in the trial.

Despite Pierre Tariot’s cautionary comment above that trial findings would not apply to populations other than that of the Colombian families without extensive further research, media reporting of comments made by other researchers, some of which were cited above, make it very clear that many think a great deal more is at stake. Benefit to the local population does not appear to be uppermost in the minds of many researchers, and the Colombians are indeed thought of by many noninvolved researchers as a “treasure trove.” This means, of course, that Francisco Lopera is poised in a very delicate position; on the one hand he must be a protector of his compatriots, among whom he has worked for decades, and on the other hand, as a participating member of the Banner Alzheimer’s Institute research team he has obligations to the team. He is not the first scientist living in a poor country to find himself in this position, but it is an extraordinarily delicate matter to negotiate.

An experienced Montréal gerontologist with whom I talked told me that, in his opinion, the proposed trial must be done. As far as he is concerned the findings will either make or break the amyloid cascade hypothesis once and for all. Not everyone agrees with this position, and many assume that even if the trial is a failure the argument will be that it should be started at a younger age or a different dosage should be used. But preliminary findings published in late 2012 (see chapter 8) suggest that this trial may indeed bring about quite a shake-up in the AD world.

By 2010 it had been shown that amyloid plaques may begin to accumulate in the brains of Colombian individuals as early as 28 years of age, and that plaque deposits increase steadily until about age 38, after which they tail off. Other research has suggested that CSF changes may be present in individuals who have a mutation for dominantly inherited Alzheimer’s as early as 18 years of age. As noted above, it has been assumed that among the Colombian families, cognitive impairment is usually readily detectable by age 45, and dementia sets in, on average, by age 51. However, a study published in The Lancet Neurology in 2011 that presented findings from the largest and longest retrospective study of predementia clinical stages in familial Alzheimer’s disease to date forced reconsideration of the age of onset of clinical symptoms in the Colombian families. Based on a sample of more than 1,700 individuals in Medellín, that included 449 carriers of the paisa mutation, it was shown that measurable clinical deterioration could be detected using cognitive testing in individuals who carry the mutation two decades before dementia onset, that is, at a median age of 35.²⁴ It is evident that changes associated with dominantly inherited Alzheimer’s, both molecular and clinical, commence earlier than had been appreciated prior to this research and, furthermore, as anticipated, the molecular changes precede cognitive changes by a good number of years.

Research that complements that taking place in Colombia is being carried out under the umbrella of the multisited consortium DIAN (Dominantly Inherited Alzheimer Network). In this consortium, cross-sectional, longitudinal studies are under way with over 260 research subjects in the United States, Great Britain, and Australia, all of whom carry one of the three genetic mutations associated with familial AD. Preliminary findings from this research, based on the tracking of biomarkers, suggest that pathological changes associated with AD may be detectable up to 20 years before the appearance of clinical symptoms of dementia. John Morris, director of the Alzheimer research group at Washington University, heads up this project. When he talked with me, he made his position clear: biomarkers “very objectively predict decline.” And he added, “People [with altered biomarkers] should be considered the real treatment target”; furthermore, this long “prodromal” phase will be very helpful in assisting drug research and the development of potential treatments.

Genetic Testing for Early-Onset AD

To date, no research has been published in which people from families that carry one of the mutations associated with early-onset AD have been asked to comment on their experiences as research subjects. Genetic testing will be a key part of the research process, but Morris insists that most people from families with familial AD with whom he has close contact, similar to the Colombian families, do not wish to know their genetic status, and they will not be informed about the results of their biomarker testing nor if they are carrying a genetic mutation for AD in the project he is conducting.

In Montréal I was able to interview just two people, brother and sister, who come from a family that carries a presenilin gene. Brenda was eager to talk; she had already been tested when I met her, and her result had been positive, although she is not clear about the name of the gene she carries. She had decided to be tested she said because, as a single woman, she wanted to have time to set her affairs in order should she test positive and arrange for appropriate help long before the AD symptoms leave her helpless. Brenda had been forced to retire at age 45 from her teaching position that she loved because her memory difficulties made it impossible to manage large classes of students studying music. As we talked it became clear that, at 52 years of age, she was resigned to a bleak future. Her father and two of her three uncles had died of the disease in their 60s, but Brenda had not been involved with caregiving because she lived in a different part of Canada at the time—that task had fallen to her mother. When I met her, she had recently undergone neuropsychological testing at a memory clinic, and said that she had really looked forward to it because she “loves tests” and had done well on the same test on previous occasions. But this time round, Brenda found that she simply could not answer the questions. As an excuse for her difficulties, she said that the kind of questions she had been asked again and again when doing these tests repeatedly in recent years now bored her. Later on in our conversation, Brenda said that she started taking the tests because her doctor had told her it was good to have a “baseline” from which to assess her condition. She finally added, hesitantly, “In the last year or so I’ve noticed that I’m really slipping.” As we talked, it became clear to me that Brenda was struggling to answer the questions I asked, and she dealt with this difficulty by taking over the conversation. I was feeling deeply saddened as we parted after talking for over an hour, and was worried about the effects our meeting might have on Brenda, although she said spontaneously that she had enjoyed our talk, and had gone out of her way to meet with me.

Brenda’s brother, Alan, lives in a different city than his sister, and they rarely communicate. Alan has not yet undergone genetic testing, having pushed the matter to one side during younger years. He has three children from his first marriage, but only after he divorced, remarried, and found himself thinking about having another child had he stopped to reflect on his family history. This time round he wanted genetic testing so that he and his present wife could “plan things well,” as he put it. When asked about his children, Alan did not elaborate on their lives other than to say that they are now adults, and it would be up to them to get tested if they so wish.

It is no doubt the case that for Colombians of Basque origin who carry the paisa mutation, termination of pregnancy should a fetus test positive for a familial AD gene would not be acceptable (I assumed, perhaps wrongly, that Alan would take a different position, given that he talked about planning “things” well). For many patients who attend the clinic run by John Morris in St. Louis, it is quite possible that pregnancy termination is also not acceptable, and this might in part account for the apparent lack of willingness to hear about the results of genetic testing that these people profess to. These are matters worthy of sensitive ethnographic investigation, particularly because examples exist from other parts of the world where genetic screening prior to marriage has proven to be an option that many individuals choose to take when families must face up to the devastating effects of single gene disorders.²⁵

An Elusive Susceptibility Gene

The APOE gene, located in humans on chromosome 19, is essential for metabolism, transportation, and processing of lipids and LDL cholesterol. This gene comes in three common, universally distributed allelic variations, APOEε2, APOEε3, and APOEε4. Each variation produces one of three different proteins, and in human populations the ε3 variant is the most common.

In 1993, a research group headed up by the geneticist Allen Roses at Duke University (who later became vice-president of genetics at GlaxoSmithKline Inc.) published articles that were the first to made an explicit association between the APOEε4 allele and increased risk for the common, late-onset form of AD.²⁶ It was shown that this allele is implicated in both sporadic cases of AD (in which there is no apparent evidence of AD in other family members, although this is often impossible to determine due to death from other causes), and in late-onset familial AD (in which other family members have clearly been affected). In the short space of time of two or three years, this association was confirmed in over 100 laboratories worldwide, on the basis of results obtained from both clinical and population samples.²⁷ These findings forced second thoughts about the received wisdom of the day—namely that all instances of Alzheimer disease in older people are “sporadic.” They also created a great deal of excitement because it was assumed by some researchers that the Alzheimer puzzle would now soon be solved.

From the outset, it was clear that the APOEε4 allele is a susceptibility gene; that is, it is neither necessary nor sufficient to cause AD. It is estimated on the basis of population studies that at least half the individuals homozygous for ε4 never get AD, and it has also been shown that between 30% and 60% of those who develop AD do not carry the ε4 allele. It was evident right away that other genes and also no doubt environmental and possibly social variables are in all likelihood implicated in disease causation. In European and North American populations its distribution is between 14% and 16%, but it is higher among Pygmies, the Khoisan, the indigenous peoples of Malaysia and Australia, Papuans, and some Native American groups—a finding to which we will return below.²⁸

Research quickly showed that when the ε4 allele is implicated in AD, the “final common pathway” resulting in plaques and tangles appears to be exactly the same as that associated with the autosomal dominant genes responsible for the early-onset form of the disease. However, the neuropathology associated with APOEε4 usually becomes manifest later in life, between the ages of the late 60s and the mid-70s,²⁹ although increasingly research has shown that these age differences are not hard and fast. It has also been shown repeatedly that the “gene dose” of APOEε4 is significant, and that individuals who are homozygous for the allele will be affected at an earlier age than those who are heterozygous, and that these individuals usually, but not always, exhibit a greater amyloid burden.³⁰ Given that somewhere between one-third and over one-half of patients diagnosed with late-onset AD do not carry the APOEε4 allele, it was apparent early on that there must be at least one other pathway, and probably several, that result in what is recognized as Alzheimer neuropathology.³¹ Today it is assumed virtually unanimously that mutually interactive genes constitute these pathways, in addition to which DNA regions with functions other than protein coding, in conjunction with environmental factors internal and/or external to the body, are all involved. These latter pathways to AD relatively rarely result in clinically diagnosable symptoms until late in life, usually after age 70 to 75, or even later, but it continues to be assumed, by perhaps the majority of researchers, that the same final common pathway is involved as that for both early-onset and APOEε4 linked AD. However, as we have seen throughout the book, these hardened assumptions are increasingly open to question as more findings are made.

Furthermore, a satisfactory explanation does not yet exist as to what sets off this train of molecular events, and basic science research increasingly makes it clear that the final common pathway is extraordinarily complex and subject to micro-molecular variation. Even so, John Hardy’s argument in a 1994 editorial in Science that genetic findings (in connection with APOE) give support to the amyloid cascade hypothesis as being causal of AD continues to be taken seriously,³² in large part because APOEε4 is associated with amyloid burden.

From the time of the discovery of the APOEε4 allele, first by means of linkage studies, followed by its mapping, it was assumed that as yet “undiscovered” genes must be implicated in AD, and gene hunting continues to be an important activity to this day.³³ Using the citation index PubMed, two neurogeneticists, Bertram and Tanzi, showed that in 2003 alone a total of 1,037 projects were carried out in which 55 genetic loci were examined on 20 different chromosomes for association with risk for AD. These authors then summarized the situation with respect to AD genetics as follows: “First, and most importantly, the heritability of AD is high. … [T]his had been demonstrated in various studies … over the past decades.” But, they go on to note: “most of the research currently being done has faulty methodology, lacks replication, and is inattentive to haplotype structure.”³⁴

The article concludes with a caveat: “while the genetic association per se [of APOEε4 with AD] has been extremely well established … there is no consensus as to how this association translates pathophysiologically,” nor, they add, how it functions “in conjunction with the other numerous candidate genes”³⁵—a statement that holds to this day.

Readers will recall that the diagnosis of AD in the majority of clinical settings is not reliably replicable and, furthermore, that very many instances of AD, probably the majority,³⁶ are never seen in clinical settings. The result is that although claims about the relationship of APOEε4 and increased risk for AD have the appearance of being very robust, considerable variation is seen from sample to sample as to just how predictive is this relationship. For example, estimates of the number of individuals with AD who carry the ε4 allele range from 30% to 90%,³⁷ and many studies do not specify whether these numbers refer to those who are hetero- or homozygous.³⁸ In addition, researchers report that between 23% and 68% of AD patients do not have the APOEε4 allele, serving to highlight the complex and elusive nature of the association between susceptibility genes and the pathology of AD.³⁹

In addition to retrospective studies of individuals who already have AD, other studies attempt to predict the number of people with APOEε4 alleles who will eventually develop AD. Considerable discrepancies exist between the estimates made in these prospective studies. Depending on the study consulted, the number of individuals who are heterozygous for the APOEε4 allele and who are expected to develop AD range from 7.6% to 47%. The range for homozygous individuals is between 21.4% and 91%.⁴⁰ The literature suggests that a person with one ε4 allele has 3 times the odds and a person with two ε4 alleles has between 8 and 30 times the chance of developing AD compared to individuals described as ε-negative.⁴¹ However, the baseline on which these probabilities are estimated is rarely provided, making such estimates questionable.

For those individuals with two APOEε3 alleles (about 60% of the population in Europe and North America), risk for AD is estimated as “average,” and it is calculated that about a quarter will develop AD once over the age of 80. Relatively few people carry APOEε2, although it is higher in some populations than others. Those who inherit two copies of this gene are thought to be at low risk of contracting AD, and ε2 appears to be protective, although two of the individuals I know who chose to have themselves privately tested because AD is present in their families found that they are carrying ε2 alleles that apparently do not protect their relatives.

One of the principal causes of confusion about genetic risk for AD is inherent to the design of much of the research. One problem is that research is often based on clinical and not population samples, thus introducing bias.⁴² When general population samples are used, the relationship between APOEε4 and AD appears to be significantly weaker than is commonly suggested.⁴³ Not surprisingly, given the complexity involved, it has been repeatedly confirmed from the latter part of the 1990s that the ε4 allele does not determine progression to dementia, and community studies have shown that up to 25% of “normal” individuals in their 80s carry an ε4 allele.⁴⁴ This matter is made yet more complex because the incidence of APOEε4 and its relationship to AD varies significantly according to the human groups under study and their geographical/environmental locations (see below). In chapter 7 ethnographic findings will be presented based on interviews conducted with individuals who have been tested for their APOE status and informed of the results of their tests.

Setting aside for one moment the question of the accuracy of reporting and of risk estimates, several researchers have argued for years that too much weight has been given to the contribution of the APOEε4 allele to AD. More than a decade ago, the biological anthropologist Alan Templeton was critical of the conclusions drawn by many researchers in connection with the significance of APOE to AD incidence. He pointed out that genomes are “commonly organized into clusters of functionally related genes,” and that APOE is part of one such large cluster. Templeton argued that when this type of gene is associated by linkage with a specific phenotype, great caution is called for, because the gene may simply be a marker for another gene or genes located nearby on the same, clustered segment of DNA.⁴⁵ It is of note that APOEε4 was first associated with increased risk for heart disease and hypoglycemia, and is implicated in yet other medical conditions (in technical jargon it exhibits pleiotropy), but its association with risk for AD has attracted the most attention in both the research world and in the media.

In 1996 in an article that was part of a three-volume issue devoted to the APOE gene published by the Annals of the New York Academy of Sciences, Allen Roses and colleagues concluded their contribution with a provocative statement. They argued that, if genotyping for APOE were to be linked with the use of other biomarkers such as neuroimaging, particularly in individuals with mild cognitive complaints, it would soon become possible to administer “anti-dementia treatments before extensive brain damage develops.”⁴⁶ This statement makes clear that a move to the prevention of AD was already being considered as a realistic option in the mid-1990s. However, in a second article published in the same volume of Annals, Roses modified his previous statement significantly, now making it clear that predictive genetic testing of “cognitively intact individuals” should not be recommended, and that APOE genotyping should be used only to confirm a diagnosis of probable AD in patients with symptoms of dementia.⁴⁷ Guidelines similar to this latter position have recently been loosened in the United States but not in the United Kingdom, Canada, or France. Even so, the new NIA–Alzheimer’s Association criteria for diagnosis discussed in chapter 4 exhibit caution and state several times that their recommendations apply to research settings alone and are not as yet for use in the clinic. In a third article published in 1998, once again in the Annals,⁴⁸ Roses, by then affiliated with Glaxo Wellcome Research and Development Unit in North Carolina, stated that a second locus for a susceptibility AD gene on chromosome 12 would shortly be announced but, despite further research, this locus was never verified.

A summary article published in 2000 by Ann Saunders, Allen Roses’s spouse and a prominent member of his research team, argued that clearly APOE is a multifunctional molecule with potential roles in amyloid deposition and clearance, microtubule stability, intracellular signaling, immune modulation, glucose metabolism, oxidative stress, inflammation, and other cellular processes. She added that the presence of an ε4 allele had been associated with poor recovery from brain trauma. Saunders also stated that the alleles of APOE differ from each other by only one amino acid substitution (an inaccurate claim; it has been known for many years that two amino acid substitutions are involved).⁴⁹ She pointed out that the differential effects on CNS and brain function are very dramatic, given these small differences among the alleles (very evident, even though two substitutions are involved).

The APOE gene has clearly presented a challenge to both researchers and clinicians from the time that its confounding role in the AD conundrum was first made apparent. But there are more findings, created largely by population geneticists and biological anthropologists, making the picture yet more convoluted. However, such findings have rarely been noted in the majority of articles written about APOE by neurologists and other clinicians, with the notable exception of a recent issue of Alzheimer’s & Dementia.⁵⁰

Human APOE—Which Allele Came First?

Since the late 1990s there has been debate among geneticists about the evolution of the APOE gene. This gene is well “conserved,” meaning that, in terms of an evolutionary time scale, it has been in existence for millions of years and is widely present in mammals, including nonhuman primates. However, other than in humans, APOE exhibits only one variant that resembles the human ε4 allele, but is not exactly like it.⁵¹ The association of this gene with deposition of amyloid plaques, neuron atrophy, and other changes in the brains of apes and certain other mammals in the latter part of their lives has been known for some time,⁵² and behavioral signs of dementia have been reported in older animals who are pets or in captivity.⁵³ It was proposed in 1988 that the ε4 allele must be the ancestral mammalian form of APOE, including that of Homo sapiens.⁵⁴

The well-known primatologist Robert Sapolsky coauthored an article in 1999 with Caleb Finch, whose specialty is the neurobiology of aging, in which they postulated a connection between the so-called grandmothering hypothesis and selection for sporadic mutations of the APOEε3 allele during the early history of Homo sapiens. The grandmothering hypothesis postulates that, as a result of the extremely long dependence of human infants after birth, it became adaptive for grandmothers during their postreproductive life to contribute extensively to child care, leaving younger women free to forage for food. Recently this hypothesis has been called into question by certain researchers, but its use to bolster a thesis about the evolution of the ε3 allele is nevertheless interesting. Finch and Sapolsky claim that if the grandmother role is important, and there is evidence from existing hunting and gathering societies that this is indeed the case, then a predisposition for dementia and cardiac disease would be selected against. This selection process may account for why the ε3 allele is by far the most common among human populations today (up to 77% among so-called Caucasians). At least one well-known biological anthropologist, Ken Weiss (personal communication), remains entirely unsatisfied with this kind of argument. It is equally possible that genetic drift rather than adaptation accounts for the increase in the ε3 allele and, given that it is estimated that currently 95% of humans carry at least one copy of ε3, it is quite possible that ε4 and ε2 will eventually in effect disappear. What we may be seeing today is an “evolutionary snapshot” of a process that will eventually drive these alleles to “extinction.”⁵⁵

However, Finch and Sapolsky, good scientists that they are, also ask if the apparently deleterious ε4 allele may confer some advantage on young adults in connection with neurodevelopment. In attempting to explain why the ε4 is more prevalent in parts of Africa and other isolated tropical regions, they note that it appears at times to confer resistance to infectious disease. The question of which APOE allele came first remains unsettled, but these ongoing arguments make it clear that, without reference to a context larger than clinical settings, knowledge about the functioning of susceptibility genes such as APOE is impoverished—genes have a deep history encompassing two time dimensions simultaneously, evolutionary and historical, and as the following chapters will make clear, the question of how exactly they are “expressed” in any one individual depends upon local environments, macro and micro, and on individual behavior.

With this in mind, the Italian geneticists Rosa Maria Corbo and Renato Scacchi carried out an extensive study in the late 1990s to examine the distribution of the APOE gene. They found that among Pygmies, the Khoisan, indigenous peoples of Malaysia and Australia, Papuans, some indigenous North American peoples, and the Lapps, the ε4 allele was proportionately the highest. Their argument is that these are groups of people whose subsistence economy was until relatively recently predominantly that of hunting and gathering and who have lived in situations were food shortages are common. Corbo and Scacchi postulate that ε4 can best be understood as a “thrifty gene,” similar to the genes associated with diabetes, in that it assists with a higher absorption of cholesterol that would have been protective in times of food scarcity. They cite James Neel, who first set out the thrifty gene hypothesis, an argument that has taken some battering lately, largely because many readers have assumed that Neel was making causal arguments about the genes associated with diabetes, which was not the case. Corbo and Scacchi, like Neel and other thoughtful geneticists,⁵⁶ make no such claims about causality, but emphasize that gene/environment interactions are at work. They suggest that the ε4 allele may have become disadvantageous only once the human life span was routinely extended due to technological and environmental changes accompanied very frequently by the adoption of a sedentary lifestyle and, particularly in recent years, the adoption of diets rich in carbohydrates and fat and low in fiber. These authors conclude that to carry APOEε4 in a “Westernized environment” may well have become disadvantageous.⁵⁷

But this is not yet the whole story. It is unusual for genes to exhibit only three common alleles, and evolutionary biologists and population geneticists believe that this apparent lack of variation should be accounted for. Several researchers have reported on an APOEε5 allele found at very low frequency in the Horn of Africa. And, in contrast to clinician researchers, biologists make it clear that many other APOE alleles in addition to the three ε alleles exist, several of which are in the regulatory area of the APOE gene and may well contribute to APOE “effects.” These alleles are relatively recent discoveries and are in effect absent in discussions about APOE and AD.

One group of molecular biologists carried out comparative research in which they looked for variation within the alleles ε2, ε3, and ε4. By examining haplotype variation (sets of single nucleotide polymorphisms [SNPs] that are transmitted together) they concluded that prior to the past 200,000 years ε4 was the only allele in existence, and therefore must be the ancestral one (even so, they take issue with the grandmother hypothesis). They also found that when they “looked beneath the surface” of these three APOE protein polymorphisms they could readily detect substantial variation (heterogeneity) in the sequencing of each of them—in other words, each allele, whether ε2, ε3, or ε4, exhibits variation depending in part on the population in which it is found and the geographical location.⁵⁸ This could account for a great many of the contradictory results that have always plagued population genetic research in connection with the APOE gene.

The Indianapolis–Ibadan Dementia project headed up by the psychiatrist Hugh Hendrie for over 17 years working together with Nigerian collaborators has shown that the incidence of AD and dementia among Yoruba are less than half that found among African Americans, even when controlling for age. However, the frequency of the ε4 allele is not significantly different in the two cohorts. Furthermore, Yoruba have a lower incidence of both vascular disease and vascular risk factors, including hypertension, than do African Americans, and, significantly, cholesterol and lipid levels are much lower among Yoruba. Hendrie’s group concludes, reasonably, that genetic and environmental factors may well both be responsible for these differences. They also note that considerable variation is likely to exist among the genomes of the studied Yorubans, as has been well established for African populations in general. As yet, they do not know how well the African American admixed population corresponds to the Yoruban population that was studied, calling for further investigation.⁵⁹ Research findings based on 338 African American subjects showed that the ε4 allele confers about the same increased risk as it does for white populations living in the United States;⁶⁰ however, the probable extent of vascular dementia among these subjects may have been a confounding factor. Clearly, examining haplotype variation, particularly in Nigeria, would be advantageous.

The methodology of a good deal of the comparative epidemiological research into APOE and AD has been criticized, almost exclusively by epidemiologists, and it appears to be virtually impossible to eliminate bias in this type of research;⁶¹ even so, the data appear sufficiently robust to draw the conclusion that both risk-reducing factors (in Africa) and risk-enhancing factors (in North America) must surely be implicated, among them other genes, their protein products, diet, environment, and possibly yet other factors.

APOEε4 and Neurodegeneration

Yadong Huang is a neuroscientist who works at the Gladstone Institute, University of California, San Francisco. He exudes excitement about his research on APOE, a topic that has occupied him for over 20 years since his time as a graduate student in China. We talked for well over an hour when I met him. He started out by saying, “I know the APOE proteins quite well.” And his concluding comment was, “I always feel good when I talk about APOE!”

Huang’s position is that, without doubt, the ε4 allele puts people at increased risk for AD due to its interaction with β-amyloid—two decades of research have demonstrated this repeatedly, he insists. But, he adds, “this is not the full picture—APOEε4 clearly has its own effect, or effects, independent of amyloid that are associated with cognitive decline and, furthermore, whatever effects the ε4 allele has, some of these changes could well take place early in life.” As an example, Huang cites a body of research showing that ε4 carriers exhibit a decrease in glucose metabolism in certain parts of their brains during their 20s. This decrease, associated with the mitochondria—the powerhouses in cells—suggests a vulnerability to AD, years before similar changes are seen in other individuals. Then Huang adds, “and there is a difference in activity that has been detected in the ‘default mode network’ [that includes the hippocampus] between people who are ε4 positive and ε4 negative.⁶² He goes on to elaborate on ε4 and its properties in association with Aβ, tau, and other molecules, and then cites animal studies that his team has carried out on the relationship of APOE and the gene known as GABA.

When I mentioned that I had been told by a number of experts that genetically modified mouse models are simply not adequate as an AD substitute for humans, Huang replies,

The mouse is a good model but we have to keep in mind that a mouse is just a mouse—they are not humans. For one thing, human APOE and mouse APOE, although they are similar, they are not the same. Of course all mice are “ε4-like,” but also they clear β-amyloid differently from humans. However, sometimes the mouse APOE behaves functionally a bit like APOEε3. This has enormous significance for clinical trials when Stages 1 and 2 are conducted in mice. And it also indicates that we shouldn’t assume the amyloid cascade is wrong on the basis of what we learn from mice. But this goes back to the point I want to emphasize: we shouldn’t put all our eggs into one basket. Right? Above all, we shouldn’t mix our efforts at drug development—where most of the money goes—with research designed to work out the disease mechanism. It’s time to think seriously about more than Aβ; we need to diversify the research.

After a pause Huang adds,

Alzheimer disease is a kind of syndrome—it’s a mixture of many subclasses. We should probably call it Alzheimer diseases. It’s like hypertension—many paths can lead to increased blood pressure. In the end there will be more than one drug for AD. … It’s too limited, just to play with the Aβ pathway. We’ve put all the money there, but we haven’t had any good luck so far. The theory, the amyloid theory, may be OK, but it’s only one of several pathways. Because of the genetics John Hardy emphasized that early onset AD and late-onset AD are caused by the same mechanism. This idea could be true, but perhaps it’s not. From accumulating evidence I’m thinking more and more that it’s not true, that it may be only part of the story. We’ve been too limited in our thinking. And, of course, we need to know why so many people who carry ε4 don’t get AD—this is a really important question that interests me greatly.

Huang has coauthored articles with his colleagues at the Gladstone Institute, one of whom is R. W. Mahley, cited above. Like Mahley, Huang is convinced that the ε4 allele is the ancestral form of APOE, and, moreover, Huang very much likes the “grandmother hypothesis” to account for selection for the ε3 allele during the course of human history. A summary of the complexity of the thinking currently made use of by Huang and his colleagues in connection with APOEε4 is made clear in a recent article; their fundamental position is that this allele is “much more than a contributing factor to neurodegeneration” due to the remarkable number of its functions critically relevant to the brain.⁶³ One striking thing to emerge from Huang’s comments is that, for him, the so-called final common pathway may not be as straightforward as has always been assumed. Although the core of the amyloid cascade hypothesis cannot be denied, a great deal of research is still called for. A position apparently also supported by John Hardy (see chapter 2) and Rudolph Tanzi, a Harvard neurogeneticist, has the following to say about amyloid:

[A]myloid is not just junk; it has a role in protecting the brain, particularly against the effects of infection. Amyloid is, in effect, part of the brain’s immune system. I’m a Baptist, for sure. We have had a knee-jerk reaction to the failure of the drug trials, but the first drugs in any trials always fail. It’s been like a 5^th grader shooting a ball from the midline on a soccer field. Now we are starting to work down the field. So far we can’t agree as to whether the problem is with an increase in Aβ levels, or whether Aβ starts to aggregate, or whether it’s to do with the ratio of Aβ42 to Aβ40—most people go for the ratio these days [Aβ42 is associated with amyloid deposition and Aβ40 appears to be protective against amyloid deposition]. (March 2010)

Tanzi added that he had published an article 20 years ago with the neurogeneticist Peter Hyslop showing how, following injury, Aβ “gums up” to protect the immune system. He insists that “we should learn from genetics,” by which he means that more effort should be spent on asking what it is that “starts the disease off,” rather than trying to “hit the clinical phenotype.” He adds, “Don’t go too far downstream; it’s better to start with the genes.” Tanzi made these comments to me as a prelude to talking about the latest technology being made use of in the world of molecular genetics—genome-wide association studies (GWAS)—to be elaborated on in the next chapter.

In 2009 when I talked with John Morris he explained that one of the projects his group was working on was to try to better understand the relationship between the APOE gene and deposition of amyloid in the brain. Using a sample of 241 individuals, the largest group to date anywhere to be repeatedly PIB scanned, resulted in the following findings:

… 75% of people who are … ε4 positive in our oldest age group and still cognitively normal are PIB-positive. Now why?! And will they dement? These are the questions that we are asking. … We call our people “participants,” not patients, not subjects, okay? We try to make them partners in our research and every year we hold a quote, “meeting,” unquote, where our researchers give back to our participants the results of the research in which they’ve been engaged. We only enroll people who indicate from the start that they understand and are willing to participate in all our studies, including the spinal tap. So, you know, the people who come are in a way pre-selected and quite committed. But we don’t divulge to them their PIB results, nor their APOE results, nor CSF results. But this is something we may need to revise in the future.

The Morris group published an article in 2010 in which findings derived from this sample of “cognitively normal individuals” age 45 to 88 years old were set out; all individuals had been genotyped and PIB scanned, and 168 had undergone CSF taps. They concluded, “[A]ge and APOE genotype interact to increase the frequency of cerebral Aβ deposition in cognitively normal older adults.”⁶⁴ The study found that Aβ deposition begins in middle age and increases in frequency with age, so that 50.0% of the individuals in the study aged 80 to 89 exhibited significant CSF changes, and in 30.3% PIB scans detected amyloid deposits in their brains. These results are comparable to an age-related frequency of neuropathological AD found by the same research group in “cognitively normal older adults” at autopsy.⁶⁵ The 2010 publication confirms that APOEε4 has a powerful dose-dependent effect on cerebral Aβ deposition (that is, two copies of the ε4 allele have a greater effect than one), but individuals who do not have ε4 alleles also exhibit age-dependent increased changes in CSF and amyloid deposition, although not to the same extent as carriers of APOEε4. This study also demonstrated, perhaps convincingly for the first time, that APOEε2 is protective against amyloid deposition; however, what is not noted in the article is that the APOEε2 allele places individuals at increased risk for hyperlipidemia.

The authors acknowledge certain weaknesses in the research protocol, but, even so, on the basis of their findings, they concluded, “Aβ is central to the initial detectable pathological changes in preclinical AD, with changes in tau likely occurring later.”⁶⁶ But, they go on, “[T]he concept of preclinical AD must remain speculative” until there is enough evidence that “cognitively normal older adults” with reduced CSF Aβ42 who are also PIB-positive, and who have other clinical indicators of preclinical AD, are at a disproportionally greater risk for developing AD than individuals without evidence of these biomarker changes. This publication concludes, “Alzheimer disease is a complex disorder and its pathogenesis almost certainly cannot be explained simply by abnormal metabolism of Aβ. However, we find powerful evidence that cerebral Aβ deposition with age is the pathobiological phenotype of APOEε4, the strongest genetic risk factor for late-onset AD. … We also find evidence that Aβ abnormalities … initiate the pathological cascade of preclinical AD.” The final comment argues that “presumptive preclinical AD” has been found in “a substantial number of cognitively normal adults”—these individuals should be tracked using longitudinal studies to “determine their risk” for Alzheimer disease.⁶⁷

Entanglements of aging and dementia, normal and pathological, continue to grip the AD world, as does profound uncertainty about the role of amyloid-β in AD causality. A 2011 summary article about the amyloid cascade hypothesis as causative of AD makes the following points, among others: It is still not clear how much amyloid-β production should be lowered, or to what extent amyloid-β clearance should be facilitated to modify the disease. Nor is it clear at what stage in the disease process modification of amyloid-β is likely to have clinical efficacy. It is also noted in this article that there is considerable heterogeneity among human brains with AD, and that it is impossible to assess accurately the quantity of amyloid deposition in the brain, especially because different parts of the brain are implicated. The authors insist that it has proved to be “very difficult to derive predictive data on the dynamics of plaque deposition,” and, furthermore, “The data … suggest that absolute levels of amyloid-β are not a key determinant in the age of onset of AD.”⁶⁸ What is more, the deposition of amyloid-β does not correlate with the presence of neurofibrillary tangles, cell loss, or dementia. These researchers nevertheless continue to support the amyloid cascade hypothesis, but ask for minor modifications so that amyloid would be understood as a “trigger.” This small change captures much better, the authors insist, the temporal separation of the complex processes that appear to take place at the molecular level over years. It is concluded that large “natural history” studies are called for in which young “normal participants” should be consistently evaluated over years. And, furthermore,

to test the amyloid-β trigger scenario is beyond the resources of a single pharmaceutical company, but if this truly reflects the AD process, then a radically different model of pharmaceutical development will need to be established to prevent the public health disaster that awaits us.⁶⁹

When two scientists who are not involved with AD research reacted to a query I raised as to why the amyloid cascade hypothesis has such staying power, their answers were revealing. One likened the situation to a horse race in which the only horse left running is lame, but even so you hang on to your bet. Another said that amyloid deposition is clearly like the outcome of a car crash, but the investigator ends up looking at the debris instead of what happened to bring about the crash—he then likened this to clearing up a car crash and assuming that will prevent future injuries.

Thomas Kuhn argued long ago that until such time as a new apparently robust hypothesis is put firmly on the table, researchers will not give up on an earlier hypothesis because of concerns about a situation bordering on chaotic, with undeniable funding implications. And clearly it cannot be flatly denied that amyloid is indeed in some way a major actor in lifelong neurological activities. Research has shown repeatedly that amyloid must in some way be implicated directly or indirectly in the causes or effects of dementia-like conditions, notably those usually diagnosed as AD. And amyloid is also undeniably involved in other related neuropathological conditions. Hence, to date, the prevailing paradigm has been tweaked several times, funding has been raised, and research proceeds apace with the cascade hypothesis as its anchor.