New Alzheimer Genes Fail for Risk Prediction, University Medical Center Study

This article was originally published here

Action Points

  • Explain to interested patients that the major risk factors for Alzheimer’s disease are age, female sex, family history, and a certain variant in the APOE gene.
  • Explain that an NIH consensus panel recently concluded that there is no hard evidence that an individual’s risk for developing Alzheimer’s disease can be modified, or that currently available therapies affect the long-term disease course. Therefore, risk prediction is valuable primarily for advance planning purposes.

The largest genome-wide association study in Alzheimer’s disease to date has identified two new genetic variants and confirmed two others, but research leaders conceded that they would have little predictive value in the clinic.

Polymorphisms near the BIN1 and EXOC3L2 genes were more common in people with Alzheimer’s disease, with odds ratios in a replication sample of 1.17 (95% CI 1.03 to 1.33) and 1.25 (95% CI 1.05 to 1.51), respectively, according to Monique Breteler, MD, PhD, of University Medical Center in Rotterdam, the Netherlands, and colleagues.

The study — actually an analysis of pooled data from one new case-control sample and several existing studies — also corroborated the association of two other polymorphisms with Alzheimer’s disease. Those variants, PICALM and CLU, were reported last year,

But Breteler and colleagues, reporting in the May 12 Journal of the American Medical Association, found that adding the PICALM and CLU variants to a model based on age, sex, and the apolipoprotein E gene (APOE) only marginally improved its ability to predict development of Alzheimer’s disease.

These genes increased the model’s accuracy by just a few tenths of one percent.

In fact, omitting APOE and basing the model only on age and sex reduced its accuracy just slightly, as measured by the area under the receiver operating characteristic (ROC) curve: to 0.826 from 0.847 in one data set, and to 0.670 from 0.702 in another. Neither was statistically significant.

If the gene discoveries have any practical significance, Breteler and colleagues wrote, it will be “in the insights they could provide for research into the pathophysiological mechanisms of Alzheimer’s disease.”

In an accompanying editorial, Nancy Pedersen, PhD, of the Karolinska Institute in Stockholm, suggested that genome-wide association studies may have exhausted their usefulness in Alzheimer’s disease.

“The next challenge is to take a step beyond gene identification and move into consideration of genetic risk in the context of environmental risk and protective factors,” she wrote.

Pedersen also argued that cluster- or pathway-based genomic analyses — that is, more focused gene hunts — may be more fruitful than additional genomewide scans for isolated polymorphisms that may contribute incrementally to Alzheimer’s risk.

Breteler and colleagues took a four-step approach to their gene hunt, starting with new data from the Cohorts for Heart and Aging Research in Genomic Epidemiology Consortium along with existing results from the Translational Genomics Research Institute and the Mayo Alzheimer’s Disease genomewide association study.

The three data sets comprised 3,006 Alzheimer’s disease cases and 14,642 control participants.

During this first step, the researchers took a first pass at identifying genomic variants that were over-represented in the cases. It yielded a total of 2,708 single nucleotide polymorphisms (SNPs) as candidates with P values of less than 0.001.

These were then examined in another data set from the European Alzheimer’s Disease Initiative, which included 2,032 cases and 5,328, with only those SNPs with P values of less than 10-5 (0.000001) selected for further analysis.

Finally, the 38 SNPs identified in step 2 were reanalyzed against data from the Genetic and Environmental Risk in Alzheimer’s Disease consortium, comprising 3,333 cases and 6,995 controls, with a P threshold of 10-8.

Only the SNPs in or near the PICALM, CLU, BIN1, and ECOC3L2 genes survived this step. Lastly, these four SNPs were replicated in yet another sample of 1,140 cases and 1,209 from Spain’s Fundacio Alzheimer Centre Educacional.

Breteler and colleagues said these genes may shed light on the etiology of Alzheimer’s disease. They are all expressed in the brain and, according to the limited functional information now available, they appear to play roles in neuronal development and/or activity.

The study also confirmed the relationship of the epsilon 4 variant of APOE (ApoE4) to Alzheimer’s disease, which had what may be the smallest P value ever seen in a medical journal: 1.04 x 10-205. Yet the strength of the statistical association failed to translate into clinically significant predictive value because of the substantial number of ApoE4 carriers who had not developed Alzheimer’s disease.

Breteler and colleagues developed predictive models based on age, sex, APOE status, and the presence or absence of the PICALM and CLU SNPs and tested them in two cohorts drawn from the general populations in the U.S. and the Netherlands. That produced areas under the ROC curve that differed by only a few percent when APOE status was included or omitted.

The addition of the PICALM and CLU SNPs to the predictive model had even less impact, increasing the areas under the ROC curve from 0.847 to 0.849 in the Rotterdam cohort and from 0.702 to 0.705 in the U.S.-based Cardiovascular Health Study participants.

These ROC values are near the lower limit of what is generally considered acceptable for risk screening purposes.

Pedersen commented that the apparently modest contribution of genetic factors to Alzheimer’s disease is surprising.

The condition, she wrote, “is one of the most heritable common, complex disorders, with a heritability of 60% to 80% and is one of the few diseases for which a single susceptibility gene [APOE] gives rise to a substantial risk.”

She suggested that clinicians and the research community should quit looking to genetics to help in risk prediction.

“Findings such as those [in the current study] reinforce the futility of using individual genetic risk profiling for Alzheimer’s disease beyond collecting information on age, sex, family history, and APOE status,” she contended.

No specific funding for the analyses was reported. Funding for collection of the underlying data came from a large number of organizations, most of which were governmental or nonprofit.

Study authors and Pedersen reported no potential conflicts of interest.

  • Reviewed by Zalman S. Agus, MD Emeritus Professor
    University of Pennsylvania School of Medicine and Dorothy Caputo, MA, RN, BC-ADM, CDE, Nurse Planner
0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply