This is Part 5 of a seven-part series on presymptomatic detection. See also Parts 1, 2, 3, 4, 6, and 7.
28 October 2009. One question that pervaded the 7th Leonard Berg Symposium held 1-2 October 2009 at Washington University in St. Louis, Missouri, is whether dominantly inherited Alzheimer disease really is the same disease as the common late-onset forms that afflict some 35 million people around the globe, according to a recent report (ARF related news story). “Can we generalize from FAD to all AD?” asked Martin Rossor of University College, London.
The question is important in part because the interest drug developers take in testing prevention in families with strong AD genetics hinges on their confidence that success in the few will translate into success in the many. In private, some industry leaders express doubt about that point, and some companies exclude eFAD families from their trials for this reason (see eFAD Research essay). That eFAD indeed models late-onset AD (LOAD) to a great extent, if not in all aspects, is a premise of the Dominantly Inherited Alzheimer Network (DIAN), and the comparison of DIAN and the Alzheimer's Disease Neuroimaging Initiative (ADNI) data may settle the issue definitively. In the meantime, scientists at the Symposium discussed what clues they have so far. See below for talks taking stock of LOAD/eFAD similarities and differences in the clinic, pathology, biochemistry, genetics, and imaging.
First, a word about nomenclature. Autosomal-dominant Alzheimer disease is the form in which a child of an affected parent faces 50-50 odds of inheriting a pathogenic mutation in either the APP or the presenilin 1 or 2 genes. It is variously referred to as “familial” or “FAD,” “early-onset” or “EOAD,” “early-onset familial” or “eFAD” (see ARF eFAD section). Strictly speaking, none of the terms is interchangeable. Familial AD often is merely a clustering, not a Mendelian autosomal-dominant inheritance pattern; early-onset AD is often sporadic in the sense that neither affected relatives, much less a gene mutation, is known. The term used in the Alzforum section on this set of aggressive forms of AD—eFAD—does not quite catch all people it means to include, either, because recently, autosomal-dominant pathogenic presenilin mutations were spotted in some rare families who get this form of AD in their seventies, i.e., with a late onset (Kauwe et al., 2007; Brickell et al., 2007). The Washington University researchers sidestepped this confusing nameology by using yet another term: “dominantly inherited Alzheimer disease.” DIAD conveniently stands opposite LOAD and echoes DIAN, making DIAN the network for the DIAD crowd.
At the Leonard Berg Symposium, speakers stressed that DIAD/eFAD overall appears to model LOAD quite closely. However, it’s no carbon copy, and for the sake of discussion, the differences got center stage for a session. DIAD is heterogeneous. The way people decline can vary a bit from person to person, said Rossor, whose center follows members of 10 families with APP mutations and 34 families with presenilin 1 mutations. Certain clinical differences do track with genetic ones. For example, people who get DIAD because their APP gene is duplicated frequently have seizures, brain hemorrhages, and white matter changes. In these extremely rare families, some affected relatives start having seizures when they are teenagers, and depression and then dementia follow within a decade thereafter.
With presenilin, some 175 different mutations are published, but clinicians have no comprehensive description of how their preclinical period unfolds. DIAN is aiming to accomplish that and in the process may match up genotypes to phenotypes. Based on what clinicians observe, it is already clear that some presenilin cases come with added signs that are atypical for AD. For example, some patients have a paralyzing weakness of the legs called spastic paraparesis, and stumbling and falls can be among the first signs mutation carriers report. One new Bulgarian family is described in the literature this month (Dintchov et al., 2009).
Corresponding to this clinical phenotype, Bill Klunk of the University of Pittsburgh Medical School, Pennsylvania, in St. Louis showed slides of DIAD/eFAD patients with spastic paraparesis who have heavy amyloid deposition in their cerebellum, an area typically spared early on in LOAD. Pathologically, a particular kind of plaque called cotton wool plaque—shaped a bit like a fuzzy ring with a hollow in the middle—appears to match up with this symptom (Dumanchin et al., 2006; Karlstrom et al., 2008), but again, there is no definitive link between particular mutations, a particular pathology, and a characteristic clinical phenotype. Myoclonus (a form of muscle twitching) and rigidity also can be part of presenilin-mutant AD, as is Lewy body pathology (Leverenz et al., 2006). Not all mutation carriers in one family will develop these additional symptoms. On the other hand, people with DIAD lack certain deficits that are sometimes seen in sporadic AD. An example Rossor gave was impairment of visual processing and is sometimes referred to as posterior cortical atrophy, or the posterior variant of AD.
In terms of pathology, DIAD can be heterogeneous as well, said Bernardino Ghetti of Indiana University Medical School in Indianapolis. Diffuse plaques, neuritic plaques, and cotton wool plaques are the major types, and vascular deposits can be extensive in some families. Other rare families have abundant pathology in their cerebellum coupled with concomitant ataxic symptoms of poor movement coordination, and yet other presenilin mutations give rise to mixed dementia/parkinsonism reflected by cotton wool plaques in both AD- and PD-vulnerable regions. Some of the most aggressive forms of DIAN, where patients die in their thirties, show large amorphous cotton wool plaques without much neuritic structure. The vasculature in these cases is also severely damaged, with large plaques breaking through vessel walls in many places, Ghetti said. Intriguingly, Ghetti noted that these cotton wool plaques contain little Aβ1-42 but a lot of N-terminally truncated forms of Aβ, either with or without pyroglutamated residues on positions glu-3 and glu-11 (Miravalle et al., 2005).
Biochemically, the heterogeneity of the 175 published presenilin 1 mutations has triggered much debate among molecular biologists about how the mutations cause AD (see ARF Davies/De Strooper Discussion; Shen/Kelleher Discussion). At this point of the debate, a frequently heard view holds that individual differences among mutations tend to have one feature in common. That is, they disturb the function of the γ-secretase enzyme complex in such a way that the enzyme generates a pathologically altered distribution of Aβ peptide variants, Bart de Strooper of the VIB Institute in Leuven, Belgium, said in St Louis. In short, the mutant enzyme makes more of Aβ long forms and fewer of its short forms (see also Part 4 of this series). The Familial Adult Children Study (FACS) aims to capture the dynamics of this process by measuring in real time the production and metabolism of Aβ in the CSF of people with DIAD. In St. Louis, FACS leader Randy Bateman noted that 26 of the planned 36 participants in this study have enrolled as of last August, and that the last nine of those have also completed all DIAN procedures.
Several speakers hinted that the most aggressive PS mutations might even express themselves developmentally. Ghetti said that families with myoclonus and seizures have ectopic neurons in the white matter, i.e., neurons that presumably migrated to the wrong place in utero (Takao et al., 2001); John Ringman of the University of California, Los Angeles, noted a trend toward less education among PS1 mutation carriers in Mexican families he studies.
“Overall, I think FAD is actually more heterogeneous than sporadic AD,” Rossor summed up. This may be counterintuitive to conventional wisdom, which loosely views LOAD as resulting from many different age-related insults and eFAD as being due to mutation of amyloid-generating genes. On the other hand, a focus on variations and extremely rare cases runs the risk of losing the forest for the trees, Bateman wrote. “In our clinical experience, the majority of DIAD families do not have seizures, myoclonus, or other “atypical” features, all of which are sometimes seen in LOAD, as well,” Bateman wrote. “We scientists love to look for differences, but the main point we agree on is that DIAD appears highly similar to LOAD overall.”
Perhaps the most significant aspect of DIAD heterogeneity is the age span at which carriers begin to get overtly ill. Presenilin mutations can become apparent from the teenage years to the seventies. Those are the extreme ends, however; most families have a 10- to 15-year spread around a mean age of onset in the forties or fifties. Besides presenting uncertainty for the carrier and practical challenges for future prevention trials, this broad range harbors some scientific promise, too, Alison Goate of Washington University said at the Leonard Berg Symposium: “There are other things besides the presenilin mutation going on that determine age of onset. They could be genetic or environmental.” If these things could be understood and exploited therapeutically, even a modest five-year delay in the age of onset (of LOAD in this case) would be a boon to public health.
Some clues on what these factors might be have already come out of research on LOAD. Genetically, the ApoE gene plays the single biggest role, with each E4 risk allele bringing down age at onset by some five years, Goate said. ApoE interacts strongly with head injury in that ApoE4 carriers are more likely to develop dementia after they sustain brain trauma. For example, boxers with dementia pugilistica tend to have E4 alleles (Jordan et al., 1997; ARF Discussion). Moreover, genetics research has generated a long list of genes that play smaller but possibly important roles in LOAD. Whether any of these genes influence age at onset in DIAD is a question DIAN could address, for example, by looking for top AlzGene risk alleles in the participating families as well. LOAD genetics just got a boost from the discovery of three new genes this past summer (see ARF related news story), but they have not been checked in DIAD/eFAD yet. How about ApoE in DIAD/eFAD? Even this oldest LOAD risk gene has not been systematically studied in eFAD families; however, it is known that in large Colombian kindreds with the presenilin 1 E280A mutation, carriers who have E4 become symptomatic at a younger age than E3 carriers, and E2 carriers stay well some years longer still. “Even in this early-onset FAD kindred, E4 influences when a carrier will get sick,” Goate said.
In terms of environmental factors, little is known in DIAD/eFAD. Research of the same Colombian pedigrees suggested, surprisingly, that people with a higher level of education were diagnosed at younger ages than less educated carriers. This was unexpected because education is thought to be protective by affording some brain reserve. In this study, the result might simply mean that more educated people were being picked up as having a problem earlier because they were performing more demanding jobs (Mejia et al., 2003). Unlike LOAD, which tends to show up in retirement, eFAD typically gets noticed first at work.
Brain imaging has advanced immensely, and numerous lines of evidence, from ADNI and elsewhere, are gradually bringing into focus a view of preclinical AD for both LOAD and DIAD/eFAD (for an update, see Part 6 of this series). Imaging has produced its share of puzzling moments, however, where it appeared to highlight the heterogeneity of eFAD. For example, in 2007, when excitement over amyloid imaging began to spread worldwide among AD research groups, Klunk, who co-developed Pittsburgh compound B (PIB) with his friend and colleague Chet Mathis, published the first PIB images of presymptomatic carriers of the C410Y and the A426P presenilin mutations. Their PIB retention began building up with an extraordinarily intense signal in the striatum, an area hit hard in Parkinson and Huntington diseases, but usually spared in AD (Klunk et al., 2007). The typical AD regions lit up, too, but much more weakly. This finding seemed at odds with the clinical and pathological picture of AD, and it prompted questions about whether eFAD truly models LOAD. “This unusual distribution was so surprising, I first accused Chet of accidentally giving me a dopaminergic agent, not PIB,” Klunk recalled in St. Louis. But PIB it was. Since then, the London researchers have observed a similar striatum-first PIB pattern in members of some of the families they follow, Rossor said in St. Louis.
In the meantime, Klunk’s group has obtained repeat scans from those and other DIAD/eFAD research volunteers. He offered this update: Over the course of four years, the PIB pattern spread out from the striatum to a typical AD pattern with increasing binding in cortical regions, whereas the original striatal binding diminished over this period of time. The pattern of PIB retention varied somewhat by mutation. Overall, Klunk said, it looks like in eFAD, amyloid deposition starts in the striatum as early as 10 years before symptoms, peaks before or around the time symptoms appear, and then spontaneously decreases. Neocortical amyloid appears later but then stays. In LOAD, amyloid deposition begins in the neocortex (in the precuneus and anterior cingulate subareas), later spreads to include the striatum, and does not decrease.
“We also see the striatal pattern of eFAD in Down’s, the only other Aβ overproduction syndrome,” Klunk said. “We think it is related to that but cannot explain it.” In the discussion, scientists noted that PIB may not see cotton wool plaques and diffuse plaques. It also does not bind non-fibrillar forms of Aβ, all of which may develop with a slightly different time course and distribution in eFAD versus LOAD.—Gabrielle Strobel.
This is Part 5 of a seven-part series on presymptomatic detection. See also Parts 1, 2, 3, 4, 6, and 7.