May 08, 2019
By Julie Gould
Lindsay Farrer, PhD, chief of Biomedical Genetics at Boston University Medical School, discusses knowledge gaps that exist between the use of genes and identifying Alzheimer disease.
Please tell us a little bit about yourself and your research interests.
Sure. I am a medical geneticist. I have been studying the genetic basis of common complex diseases—in particular, Alzheimer disease and several other very common problems. I am chief of biomedical genetics and a professor of medicine, neurology, ophthalmology, epidemiology, and biostatistics here at Boston University Schools of Medicine and Public Health. I'm also an educator, a researcher, as well as an administrator.
Prior to the study, what exactly was known about genetic variants associated with Alzheimer's disease?
Well, quite frankly, there has been a lot of research in this area. There has been tremendous advances in our understanding. Early on ‑‑ and what I mean by early on, starting in the late 1980s, early 1990s ‑‑ there were several seminal discoveries leading to the identification of genes and rare variants, i.e., mutations, in them that cause Alzheimer disease, but a very rare form.
These individuals who develop Alzheimer's have onset of symptoms in midlife ‑‑ 30s, 40s, 50s ‑‑ and the gene behaves in a typical, classic genetic pattern. In other words, if someone inherits the mutation and lives long enough, they will get the disease. People with this form of Alzheimer's represent much less than one percent of all cases in the population.
There are a few genes that have been linked to this form. One gene that encodes the amyloid precursor protein, APP, and two homologous genes named presenilin 1 and presenilin 2.
Beyond that, the biggest genetic risk factor that has been well established for more than 25 years is a gene called APOE, apolipoprotein E, that has three predominant variants. The most common one is ε3. That's the one that's most common in the population.
It's the other two variants that affect risk of Alzheimer's. The ε4 allele increases one's risk in, what we call, a dose‑dependent fashion meaning that your risk is elevated two‑ to three-fold, if you have one copy, and 10‑ to 12‑fold if you have two copies. I'm speaking particularly about people of European ancestry.
It's also known that the rarest of the three variants, ε2 is protective. People who have ε2, their risk is about 40 percent less than those who have just e3 variant.
The APOE association has been documented in many different populations, although the strength of the association is much less in African Americans and much greater among East Asian ancestry individuals compared to persons of European ancestry.
There's been a lot of work studying this gene. In the last 10 years, with the advent of high‑throughput technology using microarray chips, we've been able to interrogate the entire genome in one experiment.
As the number of people who've been studied in this way has increased, so has the number of Alzheimer’s-related genes that have been identified.
Taking into account the most recent genome-wide association study that included about 95,000 people (all of European ancestry), the aggregate number of genes identified in this way is somewhere around 22 to 25. These genes encode proteins that are involved in many different biological pathways.
Hence, there has been some enlightenment about the mechanisms leading to the disease. Some are involved in the amyloid pathway. Beta‑amyloid is one of the proteins. The abnormal form is found in brains of persons studied at autopsy with Alzheimer disease. There are also a number of genes involved in neuronal signaling and neuro development.
Many of those genes are in a pathway that involves the gene‑encoding tau protein. Abnormally processed tau protein is a major constituent of what are called neurofibrillary tangles. That is the other hallmark pathological characteristic that defines Alzheimer disease. We now know that there are Alzheimer’s-related genes involved in the lipid cholesterol pathway.
In addition, there is a major disease pathway involving inflammatory processes. There are other genes that are involved in what is called vesicular or protein trafficking, a process that moves molecules including amyloid beta in and out of cells. There are other genes whose exact role in Alzheimer’s has yet to be worked out.
I should point out that for this group of approximately two dozen genes, most of the variants that have been identified are what we call common variants, meaning they have frequencies of at least a few percent in the general population. I should also point out that individually their impact on risk is very modest compared to the ε4 variant of APOE.
These variants tend to increase risk somewhere between 8 percent‑and‑15 percent. Not a whole lot, but they do tell us about the biology of the disease. There has been another line of inquiry that has accelerated in recent years, again because of technology that allows us to sequence the entire genomes of people or the exomes, which is the coding portion of the genome.
We've identified now through the Alzheimer Disease Sequencing Project as well as other independent investigations roughly 18 to 20 additional genes. These are variants that are believed to have functional impact on the disease. Precisely how, the mechanisms have yet to be delineated. These variants are rare, mostly much less than one‑percent frequency.
They tend to have a greater impact on one's risk compared to the common variants. Again, they're very rare, so individually they don't account for much of the genetic risk in the population.
In terms of what's known generally about the genetics, there is a different genetic architecture underlying risk across populations, however across people of European ancestry, African Americans, Hispanics and Asians there appears to be certainly commonality of genetic factors that contribute to risk.
There may be other variants that are much more prevalent or relatively more prevalent in certain populations that makes them important risk factors in that group. In summary, our understanding of the genetics of Alzheimer's is greatly increased. There is still a lot of the genetic component that has yet to be worked out.
How do the rare variants that your team identified help clinicians identify AD risk? Briefly highlight what you found in this study.
Our recently published study in the journal JAMA Network Open was one of the studies carried out by the Alzheimer's Disease Sequencing Project for which I'm one of several principle investigators. This particular project is related to what I was saying before, where we have sequenced the exomes or coding portion of the genome.
That's roughly 3% of the genome. We've sequenced the exomes of approximately 6000 persons with Alzheimer disease and approximately 5000 elderly controls who don't have the disease. Most of these individuals are of European ancestry, and there is a small portion of Caribbean Hispanic ancestry.
There have been a series of studies that we published in the last several months. In the aggregate, as I mentioned before, we've identified at least 18 novel disease-related genes through the Alzheimer's Disease Sequencing Project. In the most recent study that was just published we identified several genes, using an approach that's a bit unorthodox.
In order words, it's not the typical way that scientists tend to go about finding disease associated genes. The more typical way entails comparing statistically the frequency of variants, these rare mutations, in the group of Alzheimer cases with the controls, and come up with some statistical measure signifying a significant difference. That's what gets reported.
Here, we actually focused on variants where we identified them only in persons with Alzheimer cases and not in controls. These variants, even compared to the other rare variants we've studied, tend to be exceedingly rare. Most of them have a frequency of less than half a percent in the general population, and some of them even less than a 10th of a percent in the general population.
Of the group of variants that we identified in this manner, there were two variants that were of a particular interest because they were in genes that were previously implicated in dementia. One of these genes, called TREM2, was actually one of the first Alzheimer genes identified through genomic sequencing several years ago.
It's the first and best example of using this technology. The TREM2 variant that we reported in this study is actually even rarer than the one that was initially found. The one that we identified was observed in only four or five Alzheimer cases and not controls.
What was particularly relevant or remarkable is that this variant, which has the abbreviation of Q33X (a designation indicating its location in the protein and the amino‑acids that get changed by the mutation). We realized that this variant was previously reported to be associated with a very rare form of dementia called Nasu‑Hakola disease.
That disorder is characterized by an early‑onset dementia, people in their 40s or 50s, but also accompanied by other symptoms or problems. These people in particular have bone cysts. The exact understanding of why do these two things co‑occur is not entirely understood.
Even though it's the same mutation that we identified with Alzheimer disease, which is a much more common disorder affecting people after the age of 65, the difference is that in Alzheimer disease individuals who carry this mutation had only one copy, which is usually the case with rare variants associated with risk of Alzheimer’s.
Persons with the rarer Nasu‑Hakola disease type dememtia have two copies of this rare variant. Here, the genetic difference of one versus two copies has a tremendous impact on the clinical outcome. Yes, both outcomes are dementia but dementia occurring at different stages of life and, in the case of Nasu‑Hakola disease, accompanied by bone cysts. That was one of the primary findings we reported.
The other main finding we reported is a gene that had not been directly implicated in Alzheimer's...This gene is called NOTCH3. The pathway or the function of this gene had not been previously related to Alzheimer disease, although it is structurally similar to two other genes, NOTCH1 and NOTCH2, that have been implicated in the disease.
However, mutations in NOTCH3, starting more than 20 years ago, were implicated in another rare dementing illness called CADASIL.
CADASIL, which is an acronym for a disorder characterized by cerebrovascular‑type lesions and other brain abnormalities, is what we would call a vascular‑type dementia. This is a disorder that typically occurs much earlier in life than Alzheimer disease, generally among people who are in their 40s and 50s.
What's striking here is the natural history of CADASIL, individuals who have these mutations develop severe headaches in early adulthood, in their 20s, followed years later by severe strokes, and then followed years after that by dementia.
We believe that this is a dementia has more of a vascular etiology and not the classic amyloid plaque and neurofibrillary tangle‑related disease, which is Alzheimer's. What we found that differentiates these two types of dementia is a mutation associated with Alzheimer’s that is not one of the more than two dozen NOTCH3 mutations linked to CADASIL.
What most of those mutations have in common is the gain or a loss of a particular amino acid, cysteine. The mutation that we initially identified in 10 individuals with Alzheimer disease and in no people who didn't have Alzheimer's disease involves a substitution of the amino acid alanine.
Separately, a colleague, who is studying the genetics of Alzheimer disease in families from the Utah population database, identified two other Alzheimer’s-related NOTCH3 mutations that are even rarer than the one that we identified in the Alzheimer's Disease Sequencing Project cohort.
The mutations in the Utah families also involve the gain or a loss of an alanine. What's interesting is that these three mutations are all located in a particular portion of the protein that binds another protein that also has been implicated in Alzheimer disease called JAG1.
Focusing now only on the mutation that we identified only in the 10 ADSP cases, we looked further in other cohorts with whole genome sequencing data and identified two more instances of this mutation, one in a person with Alzheimer's disease and one in a person with mild cognitive impairment which is a stage of cognitive decline that precedes or leads to Alzheimer's disease. Again, not in cognitively healthy individuals.
What's also interesting about this particular mutation, we determined after looking at the genetic profiles of all these individuals that all 10 of these individuals, the mutation occurred on a genetic background (i.e., a particular haplotype) that's not that common.
This suggests that even though these individuals are not closely related to one another, they may have a common ancestor who had this mutation. That led us to ask the question, if they all originated from a common ancestor, can we identify a particular subgroup within Caucasians or people of European heritage that they may belong to?
By doing an analysis using all the genetic data that we had on the entire sample of 11,000 subjects who had whole exome sequencing, we were able to computationally identify a particular cluster of approximately 700 persons that appears to have a particular genetic substructure to them. Of the 10 NOTCH3 mutation carriers, 8 of them were in that particular group.
That led us to consider, is there a particular subset from north, south, east, or west Europe that they may belong to? We performed another analysis looking at mitochondrial DNA. This is a separate genome within cells.
We identified two particular groups of mitochondrial DNA variants called haplogroups within roughly 30% of this this cluster of 700 individuals that coincidentally are 30 percent prevalent in people of Ashkenazi Jewish descent.
Ashkenazi Jews are those whose forefathers lived in Eastern Europe and many of whom migrated west including to the United States.
It turns out that when we looked at population databases of information with respect to this particular mutation in NOTCH3, it hasn't been observed in non‑Caucasian individuals, in other words not in people of Black African ancestry or in people of East Asian ancestry, for example.
It has been observed only rarely in people of European ancestry. Among the European ancestry subgroups, there was a particular group where the mutation was about 25 times more common and it happens to be the Ashkenazi Jews.
What we believe is that we've identified a mutation that confers a very high risk of Alzheimer's disease. It's exceedingly rare generally in the world. It's much more frequent in people of Ashkenazi Jewish descent. In fact, the prevalence of this mutation is roughly in 1 per 100 Ashkenazi Jews. It's not nothing and has high impact.
That's what we reported but more work needs to be done to validate that this has a high impact on Alzheimer’s risk and that it is truly specific to Ashkenazi Jews. That's a study that we are now embarking on. That was, again, one of the two more remarkable genes that we identified.
I should also point out one other particular finding of this study. When we looked at the genetic burden of mutations and particularly looking at all of the genes that have been previously linked to either Alzheimer disease, Alzheimer disease‑related traits, or even other dementias, roughly 95 genes, the burden of rare variants that are predicted to impact on the protein was significantly higher in people with Alzheimer's versus those without. This is within the sample of 6,000 Alzheimer cases in 5,000 controls with whole exome sequencing information.
This is something that, perhaps, is not surprising to many, however, it had never been reported before. We've demonstrated the general importance of rare variants in the risk of Alzheimer disease.
What knowledge gap still exists between the use of genes and identifying Alzheimer disease?
There are two types of gaps in knowledge. There is the gap, simply in elucidating the complete genetic architecture of Alzheimer disease. We've accounted for 30 to 40 percent ‑‑ this is just a guess, a speculation ‑‑ of the entire genetic component of Alzheimer disease.
Fishing out the rest of the genetic component, if you will, is going to require different strategies, probably because there are a variety of genetic mechanisms, and you need to do different experiments to evaluate them.
There will be continuing work doing these genome‑wide association studies of common variants in larger and larger samples that will allow us to pick up genes, again, of small effect on risk, and finding more rare variants by increasing the size of the samples that are studied using whole exome or whole genome sequencing.
That's underway, and we expect that we're going to find more genes that way. However, some of the genetic component is probably wrapped up in other mechanisms that will require other experiments.
There are probably some genes who affect or dependent upon the variation of other genes, what we call genetic modifiers or gene‑gene interaction. Similarly, there will probably be genetic factors that are influenced by non‑genetic factors, these are called gene-environment interactions.
There will also probably be genes that are behaving in very different ways that are going to require different approaches because their effect is not observable at the DNA sequence level, such as gene expression, which is the regulation of how much of a gene is transcribed into RNA and translated into protein.
There are many other experiments that are going to need to be done. That's one gap of knowledge. The other knowledge gap is going from the information about genetic variants and then linking them mechanistically to the disease process itself.
For several of these genes, we have a head start because of known biology about them. There are studies ongoing, but for some genes we don't have a clue. There's a lot of work that's going to be done, that is being done, and will be done to try to understand how these genes and, in particular, the variants that have been associated with risk actually affect the disease process.
Is there anything else you'd like to add, either about the study or the future of use of genes?
I don't have a whole lot to add, simply to reiterate and reinforce that genetics and particularly using approaches that are agnostic, meaning that we don't have a prior hypothesis, help us answer what is it that we don't know.
It's hard to ask questions about what you don't know. The genetics approach allows us to do that. There will be a continuation of that.
Secondly, these genes that we identify, not only will they help us understand disease process, but they may be actual potential targets for developing new therapies or means of more accurate diagnosis or prediction of future development of Alzheimer disease.
Patel D, Mez J, Vardarajan BN, et al. Association of Rare Coding Mutations With Alzheimer Disease and Other Dementias Among Adults of European Ancestry [published online March 29, 2019]. JAMA Netw Open. 2019;2(3):e191350. doi:10.1001/jamanetworkopen.2019.1350