Genetic forces drive a sizable portion of Alzheimer’s disease, yet only a fraction of cases thus far are explained by known mutations. A handful of recent papers used genomic sequencing to fish out new variants that, while exceedingly rare, pack a wallop in those who carry them. In the July 24 JAMA Neurology, researchers led by Margaret Pericak-Vance at the University of Miami in Florida reported that mutations in four endolysosomal transport genes boosted risk of early onset AD (EOAD). A few weeks earlier, a large collaboration of French researchers reported rare new TREM2, ABCA7, and SORL1 variants in Neurobiology of Aging, while scientists led by Henne Holstege at VU University Medical Center in Amsterdam characterized the pathogenicity of SORL1 variants and even proposed classifying this endosomal sorting protein as the fourth autosomal-dominant AD gene. A team led by Dominique Campion at University of Rouen, France, dug deep into the well-trodden territory of the three autosomal-dominant genes—APP, PS1, and PS2—and uncovered de novo pathogenic variants that cropped up in people with no family history of AD. Last but not least, Anne Rovelet-Lecrux, also at Rouen, linked a duplication in the tau gene to people with an AD diagnosis who lacked Aβ plaques.

Together, the studies move the field a step closer to filling in the missing genetic influence on AD, and could provide new targets for therapeutic strategies, commented Liana Apostolova at the University of Indiana in Bloomington. “There are more genetic risk factors in hiding that have yet to be discovered, and these studies suggest we’re on the right track,” she told Alzforum.

In the JAMA Neurology study, first author Brian Kunkle and colleagues report on their search for rare variants with large effects on AD risk. Reasoning that people with EOAD are likely carriers of damaging genetic mutations, they conducted whole-exome sequencing in 51 non-Hispanic white EOAD patients, plus 53 people from 19 Caribbean Hispanic families with EOAD; all had tested negative for known causal mutations in APP, PS1, and PS2. The scientists combed the exomes for variants predicted to have damaging effects, then attempted to validate each variant’s association with AD using exome genotyping data from a separate cohort of 1,500 EOAD patients, 7,000 LOAD patients, and 7,000 controls. Developed by the Alzheimer’s Disease Genetics Consortium (ADGC), the exome chip used to genotype this separate cohort contains more than 200,000 variants, most of which are functional, rare, single nucleotide mutations.

In their original sequencing cohort, the researchers identified mutations in known or suspected EOAD genes, including SORL1, PS1, PS2, TREM2, and MAPT. Some were known; others were new variants in genes previously tied to LOAD, including HLA-DRB1, ABCA7, and RIN3. Suspicious mutations also cropped up in genes without an AD record. A missense mutation in TCIRG1, present in a non-Hispanic white person with EOAD and segregating with EOAD in three Hispanic families, was twice as frequent in EOAD than in controls in the validation cohort. Deleterious mutations in PSD2 appeared in multiple unrelated cases in the sequencing cohort, and associated with EOAD in the validation cohort, at least when considered in the aggregate. Mutations in RIN3 and RUFY1 appeared in the discovery cohort, but their EOAD association in the validation group was nominal. Importantly, all four genes function in different parts of the endolysosomal transport pathway, which is essential for clearing cellular debris and unwanted proteins, including Aβ.

The researchers found additional rare mutations in EOAD patients, but these were not on the exome chip used for the validation cohort. For example, of 151 potentially damaging variants that appeared in the original exome sequencing cohort, only 43 were included on the exome chip.

While this filtering process allows researchers to test whether a variant is truly linked to disease, it also precludes consideration of totally new, potentially damaging mutations, said Holstege. “The mutations that are most damaging are also the most rare,” Holstege told Alzforum. “If you filter them out in this way, you quench your signal.”

Holstege took a different tack to find and classify rare SORL1 variants. Rather than filtering out undocumented variants, Holstege and colleagues made them their bread and butter. In the May 24 European Journal of Human Genetics, they reported 115 SORL1 variants from the exomes of a Dutch cohort of 640 AD cases and 1,268 controls. Fifteen of these variants were common, and not associated with AD. The other 100 were rare, occurring in less than 0.01 percent of the population. Of those, 19 were predicted to be strongly damaging, based on high scores on CADD, an algorithm that considers more than 100 variant characteristics to predict how likely a given mutation is to alter protein function or expression.

Strikingly, 16 of these suspicious variants only appeared in a single person within the entire cohort, and 14 of those had AD. None of the variants were included in prior exome genotyping studies, so the researchers could not draw upon past data to validate whether they truly associated with the disease. Instead, the researchers developed a pathogenicity scale. Weaving in data from more than 3,000 exomes sequenced separately, the researchers classified a total 181 SORL1 variants based on their combined CADD scores and rarity. They categorized those that had high CADD scores and were very rare as pathogenic. Estimated pathogenicity decreased from “likely pathogenic” to “uncertain” to “likely benign” to “benign” as variants became less damaging and more common.

The scientists found that a combination of high CADD scores and extremely low allele frequency selected out those SORL1 variants that occurred much more often in cases than in controls. The 13 variants with the highest pathogenicity resulted in truncations of SORL1, and occurred only in AD cases. The researchers predicted they would cause dominantly inherited AD, though none have yet been traced in family pedigrees.

Holstege and colleagues proposed that SORL1 take a spot alongside PS1, PS2, and APP as an autosomal-dominant AD gene. Pathogenic SORL1 mutations occurred in 2 percent of the AD cases in this study, placing them at a higher frequency than other ADAD genes. Like PS1, PS2, and APP, SORL1 protein strongly influences Aβ, as it protects APP from amyloidogenic processing and ushers Aβ to lysosomes for disposal (Sep 2007 news; Feb 2014 news). 

Classifying SORL1 as an ADAD gene would raise new questions. How to provide genetic counseling to affected families? Should mutation carriers be eligible to join the Dominantly Inherited Alzheimer’s Network (DIAN)? Clinical-grade genetic tests for SORL1 variants would be needed, a challenge developers may postpone until further data has confirmed the mutations are pathogenic, commented Apostolova. She added that while Holstege’s pathogenicity scale is an exciting tool that should be used in future studies, validation of each mutation in other cohorts, as well as functional evidence in animal and cell culture studies, should be required to elevate SORL1 to ADAD status. Rovelet-Lecrux agreed that designating SORL1 an ADAD gene will have to await discovery of multigenerational families in which SORL1 segregates with disease in an autosomal-dominant pattern. “Until we accumulate more genetic evidence, we cannot tell SORL1 mutation carriers how likely they are to develop disease,” she said.

A new study led by Rouen’s Campion and co-authored by Rovelet-Lecrux further supports pathogenicity of SORL1 variants, even if it does not provide evidence of multigenerational segregation. As reported July 13 in the Neurobiology of Aging, the researchers detected SORL1 missense and protein-truncating variants that associated strongly with early onset disease by doing whole-exome and genome sequencing of a French cohort of 852 EOAD, 927 LOAD, and 1,273 control cases. All but one of 13 protein-truncating variants occurred only in EOAD cases, and eight of 10 cases with available family information had a history of the disease. Besides SORL1, TREM2 and ABCA7 also harbored potentially damaging EOAD-associated variants in this sample. The researchers estimated that variants in these three genes accounted for 1.42, 1.17, and 1.33 percent of EOAD heritability, respectively. By comparison, ApoE4 accounted for 9.12 percent.

New Finds in Old Genes
While many pathogenic mutations in PS1, PS2, and APP have been traced in family pedigrees, additional rare variants in these established ADAD genes may yet be discovered. In search of them, researchers led by Rouen’s Campion sequenced these genes in 129 sporadic cases of early onset AD, as well as in 53 affected families. Published March 28 in PLOS Medicine, the findings included data from participants who joined the ongoing French study after 2012, when the researchers had published a similar analysis (Wallon et al., 2012). In all, first author Hélène-Marie Lanoiselée and colleagues identified 44 PS1, two PS2, and 20 APP mutations, as well as five APP duplications; 12 of the PS1 and one PS2 mutation had not been reported previously.

The most striking finding was the existence of de novo mutations in PS1. Indeed, seven of 12 new mutations occurred in sporadic cases of EOAD. In three of these mutations, the researchers were able to confirm that the carrier’s parents did not carry the new mutation. Rovelet-Lecrux, a co-author on the paper, said that the prevalence of de novo mutations in ADAD genes is likely underestimated, because routine genetic screening for these mutations is done only in familial AD cases. The de novo find underscores the importance of checking for pathogenic mutations even in patients without a family history of AD, especially people with an early age at onset, Holstege commented. Similar to the situation with rare SORL1 variants, researchers will need to decide how to categorize carriers of new and de novo mutations in established ADAD genes, she said.

Finally, results from a slightly older study led by Rovelet-Lecrux pose a different kind of classification conundrum. The authors deployed whole-exome sequencing to hunt specifically for copy number variations (CNVs) such as duplications and deletions in 335 genes predicted to influence Aβ processing, clearance, or aggregation. The researchers found CNVs in 30 out of 522 people with EOAD, but only 18 out of 584 controls. Most of these CNVs occurred in a single person in the cohort, and they included novel deletions in the PS1, ABCA7, and SLC30A3 genes previously tied to AD.

A surprising finding reared its head in four AD cases, who all had a duplication in a region of chromosome 17 including MAPT, the gene encoding none other than tau. The duplication appeared in two sporadic cases of EOAD and two with a family history. DNA available from one of those families confirmed that the duplication segregated with EOAD. Even though these four carriers had symptoms consistent with AD, the three who underwent amyloid-PET imaging had negative scans, to Rovelet-Lecrux’s surprise. All four duplication carriers had abnormal levels of CSF p-tau and tau, and three of them also had abnormal concentrations of Aβ42. The researchers also found nearly double the amount of tau mRNA in the blood of carriers than in controls.

Together, the findings suggest that despite the lack of Aβ plaques visible on PET, carriers of a tau duplication have a clinical disorder markedly similar to AD. The abnormality of CSF Aβ42 in three of the duplication carriers suggests that they could have accumulated Aβ just below the level of PET detection, a sub-threshold aggregation the researchers speculated could even be somehow caused by elevated tau.

Do these tau duplication carriers have AD? “Not if you consider Aβ accumulation as a defining feature of the disease,” said Apostolova. Indeed, in the paper, the researchers defined their disease as a tau-related dementia, proposing that it could account for a significant proportion of early onset dementia cases with no genetic explanation. While some researchers view Aβ as a mere forerunner to the more destructive tau pathology, which they consider the main event in AD, Rovelet-Lecrux shied away from separating Aβ from AD, saying that AD is ultimately diagnosed via its neuropathological hallmarks of Aβ plaques and tau tangles. She believes it will be important to screen EOAD patients without Aβ plaques for tau pathology, especially in the future once both Aβ- or tau-targeted therapies exist.—Jessica Shugart


No Available Comments

Make a Comment

To make a comment you must login or register.


News Citations

  1. Sorting Out SorLA—What Role in APP Processing, AD?
  2. SORLA Serves Up Aβ for Destruction

Mutations Citations

  1. MAPT Duplication 17q21.31

Paper Citations

  1. . The French series of autosomal dominant early onset Alzheimer's disease cases: mutation spectrum and cerebrospinal fluid biomarkers. J Alzheimers Dis. 2012 Jan 1;30(4):847-56. PubMed.

Further Reading


  1. . From Common to Rare Variants: The Genetic Component of Alzheimer Disease. Hum Hered. 2016;81(3):129-141. Epub 2016 Dec 22 PubMed.

Primary Papers

  1. . Early-Onset Alzheimer Disease and Candidate Risk Genes Involved in Endolysosomal Transport. JAMA Neurol. 2017 Sep 1;74(9):1113-1122. PubMed.
  2. . Characterization of pathogenic SORL1 genetic variants for association with Alzheimer's disease: a clinical interpretation strategy. Eur J Hum Genet. 2017 Aug;25(8):973-981. Epub 2017 May 24 PubMed.
  3. . APP, PSEN1, and PSEN2 mutations in early-onset Alzheimer disease: A genetic screening study of familial and sporadic cases. PLoS Med. 2017 Mar;14(3):e1002270. Epub 2017 Mar 28 PubMed.
  4. . 17q21.31 duplication causes prominent tau-related dementia with increased MAPT expression. Mol Psychiatry. 2016 Dec 13; PubMed.
  5. . Contribution to Alzheimer's disease risk of rare variants in TREM2, SORL1, and ABCA7 in 1779 cases and 1273 controls. Neurobiol Aging. 2017 Nov;59:220.e1-220.e9. Epub 2017 Jul 14 PubMed.