Have we learnt all we need to know from genetic studies - is genetics over in Alzheimer's disease?

Background Alzheimer's disease (AD) pathophysiology is mostly (>95%) not inherited in a Mendelian fashion. Such sporadic AD (sAD) forms do not exhibit familial aggregation and are characterized by complex genetic inheritance. Growing evidence indicates that multiple genes contribute to sAD-characteristic endophenotypes, molecular mechanisms, signaling pathways and biomarker signatures either individually or through complex gene-gene interactions, lifestyle and the environment. Discussion Under the hypothesis that low-prevalence variants showing moderate-to-high effect size may be associated with risk for sAD, two independent research groups have demonstrated that a rare variant (rs75932628, encoding a substitution of arginine by histidine at residue 47 (R47H), in the TREM2 gene, which encodes the triggering receptor expressed on myeloid cells 2) is significantly associated with an increased susceptibility to sAD. Another study has provided intriguing evidence that a low-frequency variant (rs63750847) in the APP gene is associated with a reduced risk of developing AD and a lower likelihood of age-related cognitive decline in elderly subjects without AD. Summary Recent years have witnessed tremendous development in genetics technology that has allowed full individualized genome-wide or genomic screening embracing all of the risk and protective variants for sAD, both across populations and within individuals. Hopefully, the integration of neurogenetics with systems biology and high-throughput genotyping will further pave the way to decipher all of the related causes, mechanisms, and biomarkers across the spectrum of distinct AD forms. After an almost lost apprentice decade in AD therapy development, the epoch of individualized asymptomatic screening and progress in primary and secondary prevention of sAD is probably at its dawn. Even though we are more at the beginning than at the end of sAD genetics, there is some reason for optimism given the recent identification of novel risk or protective variants (such as rare TREM2 and APP mutations) showing strong statistical associations with sAD.

investi gation of genetic variants of known patho physiological signifi cance. As a result, hundreds of genes have been tested for association with sAD. Due to a steadily increasing number of studies focusing on sAD genetics, it has become increasingly diffi cult to decipher and interpret the huge amount of available results. To address this issue, Bertram and colleagues [3] have established the AlzGene database with the aim of providing an updated, comprehensive, and unbiased online catalogue of all of the genetic variants associated with sAD, as well as updated meta-analyses to summarize the main fi ndings. According to the AlzGene database, recent large-scale GWASs [4][5][6][7][8] have identifi ed at least ten novel loci associated with an increased risk of developing sAD, that is, BIN1, CLU, ABCA7, CR1, PICALM, MS4A6A, MS4A4E, CD33, CD2AP, and EPHA1. Th e role of these genes in the pathogenesis of sAD is supported by their involvement in key pathogenetic processes in the context of neurodegeneration, including the amyloidogenic cascade, tau hyperphosphorylation, apoptotic, oxidative, cell membrane, cholesterol and lipid metabolism, and immune-infl ammatory mechanisms [9].
Existing GWAS platforms are not designed to capture rare variants which, however, are assumed to have signifi cant contributions to the heritability of sAD. It is well-known that the low frequency and the expected large number of such variants pose signifi cant challenges for study design. In this regard, a novel testing strategy, based on a weighted-sum statistic, may be useful [10]. GWASs of larger sample size will undoubtedly identify more associations and will point to additional regions in the genome for susceptibility to sAD, although over 50,000 cases and 50,000 controls may be needed [11].
Interestingly, it has recently been shown that a case control study of 5,000 cases and 5,000 controls has the power equivalent to that of a study of only 3,200 cases and 3,200 controls, or 64% of the sample size, when 20% of the case sample has been misdiagnosed [12]. Th erefore, the proportion of variance explained by SNPs may be underestimated in the presence of diagnostic misclassi fication compared with the variance explained by SNPs of the true disorder. Moreover, some genetic variation is specifi c to populations with particular continental ancestry, preventing its discovery in other populations. Popula tions of diff erent ancestry may be helpful in discovering new loci for sAD.
New breakthrough fi ndings have recently provided new impetus in the clarifi cation of the genetics mechanisms underlying the development of sAD. Under the hypothesis that low-prevalence variants showing moderateto-high eff ect size may be associated with the risk for sAD, two independent research groups have demonstrated that a rare variant (rs75932628, encoding a substitution of arginine by histidine at residue 47 (R47H), in the TREM2 gene, which encodes the triggering receptor expressed on myeloid cells 2) is signifi cantly associated with increased susceptibility to late-onset AD [13,14]. Given the well-known anti-infl ammatory role of TREM2 in the brain, it is plausible that the increased risk of lateonset AD associated with the rs75932628 variant may be due to a dysregulation of the infl ammatory processes in the central nervous system. Another study has provided intriguing evidence that a low-frequency variant (rs63750847) in the APP gene is associated with a reduced risk of developing AD and a lower likelihood of agerelated cognitive decline in elderly subjects without AD. Th is rare polymorphism results in an alanine-tothreonine substitution at position 673 of the APP protein (A673T) [15]. Th e close proximity of the A673T polymorphism to the proteolytic site of the beta-site APPcleaving enzyme 1 (BACE1) suggests that this variant may result in impaired cleavage of APP by BACE1 in subjects bearing the A673T variant. Th e discovery that a genetically determined reduction in the production of the amyloid beta peptide provides dramatic protection against the development of sAD supports the clinical usefulness of the current amyloid-directed therapeutic research eff orts. Th erefore, these genetic fi ndings support the hypothesis that the failures of recent phase III clinical trials targeting the amyloid beta peptide may be due, at least in part, to the late timing of intervention.

Discussion
What can we expect for the future of AD genetics? Th e recent development of high-throughput next-generation DNA sequencing technologies will surely play a paramount role in screening the whole genome and identifying novel genetic variants infl uencing the risk of sAD [16]. Important initiatives like the 1000 Genomes Project [17] -funded by the US National Human Genome Research Institute consortium -have already made impor tant progress toward this aim. Hopefully, future technical improvements will open new horizons for improved assessment of the genetic susceptibility to sAD, a better characterization of its endophenotypes, and the study of pharmacogenomics of drug response in sAD patients [18]. Besides traditional genetics, high-throughput next-generation sequencing technologies may also be involved in novel discoveries in the fi eld of sAD epigenetics. For example, chromatin immunoprecipitation (ChIP) combined with DNA microarray (ChIP-chip) has been successfully utilized for the study of protein-DNA interactions. In addition, ChIP followed by sequencing (ChIP-Seq) technology might allow the study of posttranslational modifi cations of histones and the location of transcription factors at the whole-genome level. Moreover, methylated DNA immunoprecipitation (MeDIP) may be useful for unbiased detection and characterization of DNA methylation patterns [9]. Another potential strategy to shed more light on the multifaceted complexity of sAD is systems biology, an innovative multilevel paradigm. By using systems biology, structurally and functionally diff erent biomolecules may be simultaneously measured over time in networks of cells or even in whole organisms. Th is strategy can allow the characterization and integration of distinct disease endophenotypes, as well as the study of common features shared by diff erent neurodegenerative disorders [19].
Generally, it is diffi cult for clinical case-control series to identify genetic risk factors for sAD based on clinical diagnosis alone. Th e use of objective and highly reproducible brain system endophenotypes can make it easier to identify sAD genetic risk factors and to under stand their impact on brain systems. Established genetic risk factors for sAD can then subsequently be studied for their infl uence on the speed of disease progression.
We argue that autosomal dominantly inherited AD (ADAD) and sAD subtypes may represent distinct entities that may be diff erent from both the genetic and pathophysiological standpoints. Such genetic diff erences could in turn be associated with distinct and specifi c multimodal biomarker signatures, possibly requiring diff erent thera peutic strategies. In this context, the fi rst cross-sectional biomarker study supported by the Dominantly Inherited Alzheimer Network [20] and the upcoming studies of disease-modifying therapies in asymptomatic mutation carriers may represent important hypothesis-testing and hypothesis-generating milestones that could accelerate a shift from the traditional conceptualization of monogenic FAD to the novel complex non-linear dynamic sAD model [2].
Finally, the use of polygenic risk scores may theoretically be useful for the prediction of certain complex diseases like sAD. Th e approach has been based on the contribution of counting multiple alleles associated with disease across independent loci. Whether polygenic risk scores may assist in the prediction of risk of sAD is unknown and should be addressed in future studies.

Summary
Recent years have been characterized by remarkable progress in genetics/epigenetics technology that enabled full, individualized genome-wide or genomic screening covering all of the risk and protective variants for sAD, both across populations and within individuals. Th e integration of neurogenetics with a systems biologybased approach and the use of high-throughput genotyping are expected to untangle all of the related causes, molecular mechanisms, signaling pathways and biomarkers throughout the spectrum of distinct AD forms. Th e era of individualized asymptomatic screening and therapy development in primary/secondary prevention of sAD is undoubtedly at its birth. Although we are more at the beginning than at the end of sAD genetics, optimism prevails given the recent important characterization of novel risk or protective variants (including rare TREM2 and APP mutations) displaying signifi cant statistical associations with sAD.

Competing interests
The authors have no competing interests to declare.