Variation in the length of a poly T section within intron 6 of Tomm40 can improve the prediction of when a given person will likely develop AD, Roses claimed. “Basically you can take anyone who is ApoE3/3, 3/4, or 4/4, who make up 85 percent of AD cases, and we can assess their risk [of AD], based on genotype, for the next 5-7 years,” Roses told ARF.

Roses said that in consultation with the FDA, he has devised a clinical trial that will not only validate whether Tomm40 analysis can predict age of onset of AD, but based on that analysis also stratify subjects for therapeutic intervention to delay onset. Steve Ferris, New York University, was at the CTAD meeting and said that he liked the trial design. “It’s a tremendous shortcut to combine risk prediction with age of onset, and combine that with a clinical trial of treatment,” he said. The trial is planned to begin enrolling at Duke, a site in Russia, and two sites in Australia next year.

Roses’s Tomm40 analysis relies on phylogenetics, which is usually the study of evolutionary relationships between different species, genera, and even phyla (hence the name). Roses and colleagues used this analysis to construct a family tree of a 10 Kb sequence of DNA within the ApoE/Tomm40 region of chromosome 19. Over the course of human evolution, multiple mutations have occurred at this locus, some on DNA strands carrying ApoE4 and some linked to ApoE3. The analysis can determine which mutations in the region came first (the base of the tree) and which ones came later (the branches). It turns out that the tree’s branches, or clades in phylogenetic parlance, can be separated into several major groups based on the length of the poly T variation. Because these mutations are in linkage disequilibrium with the ApoE allele, some of the poly T variants are more often associated with ApoE3 and some with ApoE4. The evolutionary history of this region of DNA is, in fact, more complex, with some mutations occurring more than once, sometimes on strands carrying ApoE3 and sometimes linked to ApoE4.

Nevertheless, the phylogenetic analysis shows one branch point in the history of this region of DNA that results in two major clades, one carrying long forms of the Tomm40 poly T mutation (19-36 T) and one carrying short forms (14-16 T). While ApoE4 is predominantly (98 percent of the time) associated with the long forms, ApoE3 can be linked to either short or long poly Ts. Essentially, anyone who has an ApoE3/4 genotype has one long Tomm40 poly T (on the ApoE4 strand) and either a short or long poly T on the ApoE3 strand.

How, then, does this relate to Alzheimer disease? “We looked at the age of onset in patients, most of whom had autopsy-proven Alzheimer disease, who were ApoE3/4. All the long/long Tomm40 variants came out the same as ApoE4/ApoE4 data previously, with average age of onset of about 70. But when we looked at the short/longs—short attached to ApoE3 and long attached to ApoE4—age of onset came out to 78,” said Roses. In essence, the length of the Tomm40 poly T variants determines age of onset in ApoE3/4 individuals. If replicated, the eight-year difference would be dramatic, given the suggestion that the incidence of AD could be halved if the age of onset were pushed out by five years. The number of people with AD is slated to reach 115 million worldwide by 2050 if present trends continue (see World Alzheimer Report [.pdf]). Roses cautioned that the phylogenetic data applied only to Caucasians and that other ethnic groups may have a different Tomm40/ApoE evolutionary history.

The situation gets more complex when a person is ApoE3/3 because the Tomm40 poly T possibilities are threefold; short/short, short/long, or long/long. “When we look at [ApoE3/3s] we see age-of-onset curves that cover the whole spectrum from about age 55 through 90,” said Roses. Within that spectrum, Tomm40 creates a pattern, however. People who are ApoE3/3 and have two very long copies of the poly T repeat may have an earlier age of onset, before age 70. “And when we looked at a small group of patients who have Alzheimer disease that began in their fifties, several of those are not just ApoE4/4. There were also ApoE3/3 and ApoE3/4 patients, with a high frequency of ApoE3 with a very long Tomm40,” said Roses. The data help explain why some people with ApoE3/3 are at higher risk for AD. It may also explain why some people who are ApoE4/4 are spared. In their study sample of 255 people, Roses and colleagues found two rare cases who carried ApoE4/4 but were heterogeneous for the Tomm40 poly T, having one short and one long form. Both those individuals had a recorded age of onset of 78, said Roses, which equals that of people who possess an ApoE3/short;ApoE4/long genotype. The data are slated for publication in November in The Parmacogenomics Journal.

“It remains to be seen whether or not Tomm40 turns out to be a true association with age at onset or not, or just linked to ApoE4,” said Rachelle Doody, Baylor College of Medicine, Houston, Texas. “But meanwhile, it is an intriguing hypothesis that Alzheimer’s might somehow be related to the transport of proteins across the mitochondrial membrane,” she added. Doody has worked with Medivation Inc., a company that sponsors clinical and preclinical studies of Dimebon, which showed some benefit in AD patients and is purported to have a mitochondrial mechanism of action (see ARF related news story). “Medivation has been saying for some time that the mitochondrial permeability transition pore and flux of Aβ across the mitochondrial membrane might be part of the physiology of the disease,” said Doody. “So Tomm40 is another little convergent piece of data that hasn’t been replicated yet.”

The Tomm40 phylogenetic data address the fundamental question of which polymorphisms on the ApoE stretch of DNA are actually contributing to the risk for AD. Some studies, especially genomewide association studies (GWAS), do not sequence entire regions of DNA but instead rely on single nucleotide polymorphisms, or tag SNPs (see ARF related news story) that lie nearby. In the case of ApoE, the DNA chips that are used to simultaneously measure ~1,000,000 polymorphisms in GWAS studies are actually not measuring ApoE directly, but are measuring the nearby Tomm40 and ApoC1 genes, because that is where the tag SNPs lie. In fact, recent GWAS analysis turned up three new potential risk genes for Alzheimer disease, each with statistical p values of around 10-8 to 10-7 (see ARF related news story). People made a big deal about those, but not about several SNPs in the Tomm40 region with extraordinarily high p values, one of 10-157, said Roses. “They ignored it because they thought it was explained by ApoE, but ApoE wasn’t even on the chips they were measuring.” Roses and colleagues originally found the Tomm40 polymorphisms by deep-sequencing the locus on chromosome 19, rather than relying on detection by DNA chips. “We found that the p value really depended on the allele frequency differences. No matter how we analyzed it, ApoE could not account for the p values we had,” he told ARF. That’s when they dug deeper and found the poly T mutations.

Roses, who —would you guess?—is partial to red wine, has set up two virtual companies to commercialize research and development of this data. Shiraz Pharmaceuticals Inc. will manage intellectual property derived from the Tomm40 discovery, while Zinfandel Pharmaceuticals Inc. will carry out risk validation. That would be a five-year study, noted Roses. Because he does not want to wait for validation of the Tomm40 test before running a prevention trial, he worked with the FDA on a Voluntary Exploratory Data Submission to approve a simultaneous validation/prevention study. “On October 7 we got our answer and it was consistent with one of the scenarios we discussed” he said (see Opportunity for Prevention of Alzheimer’s study, or OPAL).

If it turns out that the length of the Tomm40 poly T does have a dramatic impact on age of onset, then that could call into question all previous studies that have stratified analysis based on ApoE4, suggested Lon Schneider of University of Southern California, Los Angeles. “From a pragmatic clinical level, this work suggests that the wrong Tomm40 polymorph raises the risk level of ApoE3s to the same level as 4s. This implies that all past research comparing E4s to E3s is muddied because it's like comparing E4s to a mixed group in which about half have the same risk as E4 carriers and half have a lower risk,” Schneider told ARF. He added that he hopes prior studies saved DNA samples.

When nudged at CTAD to say what treatment might be used for the intervention part of the OPAL study, Roses hinted that it may be a PPAR-γ agonist, such as rosiglitazone. There is evidence that PPAR-γ agonists can improve mitochondrial function (see Wu et al., 2009), increase neuronal spine density (see ARF related news story), protect neurons and synapses against amyloid-β (see ARF related news story), and improve learning and memory in mouse models of AD (see Escribano et al., 2009). The drug also has a good safety profile. “If you are going to give a drug to normal people to prevent something, the FDA would like it to be the safest possible drug,” said Roses.

However, several Phase 3 clinical trials of rosiglitazone in AD conducted by GlaxoSmithKline (see ARF related news story) showed no significant effects by E4 or E3 status. “Perhaps GSK should genotype the saved DNA for Tomm40 polymorphs and reanalyze the efficacy outcomes based on both ApoE and Tomm40 status,” Schneider suggested. Roses said the company had turned down a request to this effect. For her part, Doody questioned the value of such post-hoc analysis. “If Tomm40 does get replicated and proven, then it will be of interest as a potential modifier, but that would not automatically lead to subgroups in clinical trials because age at onset isn’t the factor of importance. It is whether you have the disease and whether you can be treated with the agent,” she said.—Tom Fagan.