Jonathan L Haines. The Journal of Law, Medicine, and Ethics. Volume 46, Issue 3, 2018.
Alzheimer disease (AD) is a huge and growing societal problem with upwards of 35% of the population over the age of 80 developing the disease. AD results in a loss of memory, the ability to make reasoned and sound decisions, and ultimately the inability to take care of oneself. AD has an impact not only on the sufferer, but their caretakers and loved ones, who must take on a costly and time-consuming burden of care. AD is found in virtually all racial and ethnic groups. Genetic influences on AD are substantial, and there has been a 30 year history of both success and failure. Mutations for rare early onset forms of the disease have been identified, but this information has not yet led to an effective treatment. Multiple common genetic variations have also been identified, and have led to new insights into the potential role of microglia cells in addition to neuronal cells in the brain. Despite intensive efforts, a significant portion of the genetic etiology of AD remains unknown and must be identified.
Epidemiology of Alzheimer Disease
Alzheimer disease (AD) is a progressive neurode-generative disorder and is the most common form of dementia occurring in >35% of individuals age 85+. It affects more than 5 million individuals in the U.S. and is the 6th most common cause of death. The clinical hallmarks of AD are slowly progressive memory loss and alterations of higher intellectual function and cognitive abilities. Pathologically, AD is characterized in the brain by neurofibrillary tangles in the neurons of the cerebral cortex and hippocampus, and the deposition of amyloid within senile plaques and cerebral blood vessels. A definitive diagnosis of AD can only occur after death with direct examination of the brain. Clinical and neuropsychological testing in tertiary care clinics result in a correct diagnosis ~90% of the time. Medications only marginally and temporarily affect disease severity and progression. In the intermediate and final stages, individuals with AD require complete care, extending the social and financial burden to caregivers and family members. This burden will only increase as the prevalence of AD is estimated to triple by 2050 with a resulting cost of >$50 billion per year. Accurate pre-symptomatic prediction of disease is not possible and preventive measures will only become available when the pathophysiology underlying AD is much better understood.
The underlying causes of AD remain largely unknown. Environmental risk factors include physical trauma (typically head injury), lower educational attainment, and lifestyle choices including the often-related risk factors of a high fat diet, obesity, type II diabetes, and cardiovascular disease. Although AD occurs in all racial and ethnic populations, it remains unresolved if the incidence and prevalence differ significantly across the groups. Rates do differ by sex, with women having a higher rate of AD than men. Age is perhaps the strongest risk factor of all, with prevalence increasing rapidly between age 65 and 85. After age 85, up to 35% of individuals will have AD.
Genetic Epidemiology of Alzheimer Disease
Although AD was first described in 1907, it was not until the 1950s that any suggestion of a genetic influence was made, and reiterated in the early 1980s. Since that time, numerous studies have demonstrated that AD is highly heritable (~80%) but is also genetically complex. Monozygotic twins are more frequently concordant for AD than dizygotic twins, and a family history of dementia is still the most important risk factor for AD, elevating the risk up to 5-fold. With the exception of rare early onset families, however, AD inheritance is not consistent with fully causative rare variants. Instead, the many large multi-case families having late-onset AD suggest a more complex combination of common, rare, and private genetic variants varying in effect size.
Early Success in Understanding the Genetic Underpinnings of AD
It was noted in the 1960s that a few families existed where the onset of dementia was unusually early (typically between ages 40-60), and the inheritance pattern was consistent with a single causative mutation. The search for this mutation began in earnest in the 1980s as genetic technologies began to unlock the mysteries of the human genome. The first of these mutations to be identified in 1991 was in the amyloid precursor protein (APP), which codes for the protein that aggregates into the hallmark amyloid plaques in the brain. Studies of additional families quickly demonstrated that there were other rare mutations, and by 1995 mutations in two additional genes, in the Presenilin (PSEN1) and Presenilin 2 (PSEN2) genes were identified. Although these were major breakthroughs and helped to uncover some of the pathophysiology of AD, in aggregate the mutations in these genes explain less than 1% of all of AD.
The much more common late-onset AD (LOAD; onset > age 60) presents a more difficult genetic puzzle. While there are many families with multiple individuals with AD, the late onset (often age 70s-80s) and competing causes of death make it hard to trace the disease through a family, thus compromising most of the available statistical genetic tools. Fortunately, by modifying and combining these tools, the first gene for LOAD was identified in 1993. The apolipoprotein E (APOE) gene has three major alleles (-2, -3, -4), with APOE-3 being by far the most common. The APOE-4 allele imparts a significantly elevated risk of AD (2-3 fold for 1 copy and 6-12-fold for two copies). In contrast, the APOE-2 allele imparts a significantly decreased risk of AD (2-fold reduction). Current estimates suggest that the genetic variation in APOE may explain as much as 25% of the genetic causes of AD. However, more than half of the AD cases do not have the high-risk E4 allele and the population attributable risk of AD due to APOE may be as low as 10% to 15%. Of particular interest, the effect of the APOE variants differs by ethnic background, with individuals of African ancestry having an attenuated risk and those with an Asian background having an elevated risk.
Identifying Additional Genetic Effects
To identify additional genetic effects in AD, it is necessary to take advantage of the rapid technological progress in genomics. The advent of very dense genotyping “chips,” which measure up to 2,000,000 common DNA variants across the genome, made genomic examination of very large case-control datasets possible (generally called genome-wide association studies, or GWAS). By carefully applying detailed statistical methods to these data, samples from around the world have been tested. Currently there are 19 loci significantly associated with the risk of AD, but these loci have much smaller allelic effect sizes (odds ratios 1.1-1.2) compared to APOE (odds ratio ~2-3). Despite these intensive efforts, in total it is likely that <50% of the heritability of AD has been explained.
The 19 loci are all defined by common genetic variation (having >5% frequency in the population). It has been hypothesized that rarer variants (having <5% frequency in the population) may also contribute to the remaining genetic variance in AD. In part this was already known, as the early studies of large pedigrees with early onset AD demonstrated this power by identifying the Amyloid Precursor Protein (APP) and the Presenilin 1 and Presenilin 2 (PS1 and PS2) genes. APP, PS1 and PS2 all suggest a common pathway for the aggregation and generation of the Aβ. The hypothesis is that similar rare variations might also explain some of the more common LOAD. Because the population-scale GWAS studies of cases and controls are under-powered to detect low frequency variants, pedigree-based studies, which provide a powerful enrichment strategy for rare variants, are an attractive alternative. These rare variants are unlikely to be causative in the usual sense, but would likely significantly raise the risk of developing AD, much like the APOE-4 allele. If sufficient numbers of the rare risk variants exist, they could explain an important fraction of the “missing heritability.” The Alzhiemer Disease Sequencing Project, a large national effort to understand the genetic architecture of AD, has collected and DNA sequenced over 100 such pedigrees, and clues to the location of such loci have recently been identified.
Because the population-scale GWAS studies of cases and controls are underpowered to detect low frequency variants, pedigree-based studies, which provide a powerful enrichment strategy for rare variants, are an attractive alternative. These rare variants are unlikely to be causative in the usual sense, but would likely significantly raise the risk of developing AD, much like the APOE-4 allele. If sufficient numbers of the rare risk variants exist, they could explain an important fraction of the “missing heritability.”
An alternative approach toward using genetics to understand the pathobiology of AD is to look not for variants that increase the risk of AD, but to look for variants that protect against AD. The heavy focus on finding AD risk variation is understandable as it can lead directly to understanding causation. However, such a focus may have blinded us from examining an equally important alternative that heritable variation may protect against AD. Such variation can illuminate biology, but it can also identify targets for therapies, as has recently been done by identifying the role of PCSK9 in lowering lipid levels.
We already know that the APOE-2 allele confers protection from developing AD and that individuals carrying an APOE-2 allele and an APOE-4 allele have a substantially attenuated risk and later onset. More recently, rare variations in other genes (e.g., ABCA7, REST, TREM2) have been identified that provide protection from AD. The underlying study design for finding protective variants is to identify individuals who are at high risk of developing AD but remain cognitively normal. This can be done in several ways. One example is to identify individuals who are APOE-4/4 homozygotes (who have up to 12-fold excess risk) who have aged past the usual age at onset, and compare them to APOE-4/4 homozygotes who have developed disease. A second approach is to identify individuals with a strong family history of AD and who have a high genetic load. The genetic load can be measured by counting the number of AD risk alleles they carry at the 19 known AD loci, and summing this into a genetic risk score. Isolated populations, such as the Amish, who have large families and excellent genealogies, are ideal for such a purpose.
There have been numerous investigations of traits related to AD, either biomarkers (if they are physiological measurements) or endophenotypes (if they are more qualitative measures). The hypothesis is that these measures, being both easier to obtain and more directly representative of the pathological processes, will have stronger genetic signals and perhaps more predictive ability. One area of particular interest is neuroimaging, either by magnetic resonance imaging (MRI) or positron emission tomography (PET). Both can provide detailed images of the living brain and identify changes representative of AD. Numerous studies have found correlations between various imaging measures and AD, although the correlation of early findings to the eventual development of AD is still controversial.
There have been numerous investigations of traits related to AD, either biomarkers (if they are physiological measurements) or endophenotypes (if they are more qualitative measures). The hypothesis is that these measures, being both easier to obtain and more directly representative of the pathological processes, will have stronger genetic signals and perhaps more predictive ability. One area of particular interest is neuroimaging, either by magnetic resonance imaging (MRI) or positron emission tomography (PET). Both can provide detailed images of the living brain and identify changes representative of AD.
A second approach is examination of biomarkers in the cerebral spinal fluid. Measurements of the ratio of the Aβ40 and Aβ42 ratio has proven predictive of the development of AD, but such measurements are still somewhat invasive and expensive and are not routinely performed.
Therapeutic Implications of AD Genomics
There was significant optimism when the first genetic mutations for AD were identified that this would lead quickly to therapeutic treatment. Much has been learned about the amyloid cascade and its role in AD, in particular the role that the abnormal Aβ protein plays. However, all the clinical trials so far where some aspect of amyloid processing has been targeted have failed. By examining the potential interactions of the additional genes now identified as risk factors for AD, a somewhat different picture is emerging that suggests an important role for immunity and inflammation, and the importance of microglia in the AD process. Clinical trials currently being carried out include the immune response-based therapies targeting Aβ and γ-secretase (Presenilin) inhibitors targeting the production of Aβ, and many more are in the pipeline.
Summary
Alzheimer disease is one of the major scourges of old age and is a growing societal problem. Genetic studies have identified numerous risk and a few protective factors for AD. This information illuminated the amyloid hypothesis as a major driver of AD pathology, and has led to many clinical trials, which unfortunately have failed. The most recent genetic data suggests that alternative pathways, including the immune system may be important and should be explored. These findings also suggest that other cell types besides neurons (e.g., microglia) may have equally important roles. With the advancing technology of genome sequencing, identifying new genetic risk and protective variants can be applied to ever larger datasets of cases, controls, and families. Much has been learned but much remains to be learned.