. Apolipoprotein E Genotype and Sex Risk Factors for Alzheimer Disease: A Meta-analysis. JAMA Neurol. 2017 Oct 1;74(10):1178-1189. PubMed.


Please login to recommend the paper.


  1. Surprisingly, the effects of sex on ApoE4-related risk of Alzheimer's disease have not been well examined. The studies most referenced are from the 1990s, despite the huge growth in information about ApoE genotype and AD since then. Now, in a meta-analysis of 27 studies and over 30,000 individuals, Neu et al. show that, grouping all ages together, there is no difference in ApoE4-related risk in women than in men: ApoE3/4 women, Odds Ratio=3.1; ApoE3/4 men, OR=3.3. Logistic regression analysis did show that the ApoE4 effect occurs earlier in women, resulting in a higher OR in ApoE3/4 women than men in the age group 65 to 75 (4.4 versus 3.1). This finding supports the earlier, smaller studies that ApoE4 affected women more than men, but this effect is apparently not nearly as strong as previously thought.

    There were other findings here that help define the risk associated with ApoE2. ApoE2 was associated with decreased risk of AD (as expected), but ApoE2 women were protected better than men (OR = 0.5 compared to 0.7). Female and male ApoE2/4 individuals had an increased risk of AD (OR=2.1-2.3), consistent with the view that ApoE4 has a stronger negative risk than ApoE2 has a protective risk. In both sexes, ApoE genotype had a much smaller effect on MCI, perhaps because non-AD patients can be found more in this group. No sex effect on ApoE4-related conversion from MCI to AD was observed. Together, these data help define a broader set of individuals who would benefit from any ApoE4-related therapeutic approaches.

    View all comments by G. William Rebeck
  2. In this meta-analysis of data from nearly 58,000 cases and controls, the authors report no overall sex differences in clinical AD risk in ApoE4 heterozygotes or homozygotes 55–85 years of age. But they did see higher odds ratios in women at younger ages (i.e., 65–75). Furthermore, they found that the ApoE2 allele had a greater protective effect in ApoE4 carriers than noncarriers. While one cannot exclude an impact of ascertainment bias on this primarily retrospective case-control study, I find it interesting and important for several reasons:

    It reminds us that sex differences are relevant to the study of Alzheimer’s disease, and that any differences might provide clues regarding disease mechanisms and targets at which to aim treatments.

    It suggests the value in incorporating information about sex in prevention trials, when relevant. For instance, sex may be important for sample-size estimates for those prevention studies that use cognitive or clinical endpoints in younger persons with the ApoE3/4 genotype and to ensure comparable proportions of males and females in the active and placebo treatment groups. On the other hand, sex may be less important for ApoE4 homozygotes or those with the ApoE2/4 genotype, and its implications for prevention trials in younger ApoE4 carriers using biomarker endpoints needs to be clarified.

    It highlights the value of the rapidly growing Global Alzheimer’s Association Interactive Network (GAAIN), permitting researchers to initially survey and then analyze data from the largest number of relevant subjects. This federated data resource—and shared data from prospective cohort and case-control studies in general—will have great value for the entire field.​

    View all comments by Eric M. Reiman
  3. This new study has about twice as many cases and the same number of controls as our 1997 study. The main finding of the new paper showing increased risk of AD among women between ages 65 and 75 is identical to what we reported for that same age period. The notable difference between studies is that the portion of our sample on which the sex-difference in risk was based was among clinic- and autopsy-based samples, which tend to be younger than subjects ascertained in community-based samples.

    Community-based data sets were included in their study, whereas we had such samples but excluded them from this specific analysis because of ascertainment bias. This probably explains why we saw an attenuated finding at even earlier ages (as low as age 50) but did not observe a diminishing effect of the sex difference at later age. It would be interesting to see what they find only among the clinic- and autopsy-based portion of their sample. Take-home message (for me) is that it is nice to have our findings confirmed 20 years later. ​

    View all comments by Lindsay Farrer

Make a Comment

To make a comment you must login or register.