Edited by: Leonard C. Schalkwyk, University of Essex, United Kingdom
Reviewed by: Alexandra M. Lopes, University of Porto, Portugal; Emma Meaburn, Birkbeck University of London, United Kingdom
*Correspondence: Khyobeni Mozhui
This article was submitted to Neurogenomics, a section of the journal Frontiers in Genetics
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Genes encoding mitochondrial ribosomal proteins (
Aging is a complex biological process that is characterized by overall decline in health, vigor, cognition, and increased vulnerability to numerous diseases. In addition to lifestyle and environmental factors, molecular pathways and cellular processes that are conserved across species may contribute to aging (Kenyon,
Lifespan and mortality, while related to aging, are not definitive measures of aging and can only capture a gross outcome. There are many other phenotypes that define the health, vigor, and functional fitness of individuals as they age. The genetic and phenotypic heterogeneity partly explains why only a few gene variants have been consistently associated with aging. An exceptional case is
Here we use the extensive resources from the Women's Health Initiative Memory Study (WHIMS) to test the collective effect of the
The multi-center WHI study was launched in 1993 (
The first aging trait is the gross outcome: all-cause mortality or survival time. This was measured as days from enrollment to participant's uncensored death, the last National Death Index date, or censored at end-of-follow up (Seguin et al.,
For survival time, we constructed Kaplan-Meier survival curve for all participants. After removing 14 cases with missing data, the survival function was computed for 4,490 samples with 1,282 deaths. A Cox proportional hazards regression with adjustment for age was performed to evaluate baseline predictors of survival time. Only factors that were significant from this analysis were included as covariates in the genetic association test. Survival analysis was done using the “survival” R package (Therneau and Grambsch,
For cognitive aging, we limited the analysis to 4,284 participants who had repeated measures of the 3MSE exam from baseline and at least 2 follow-up visits. For this subset, the retention rate is high and over 85% of the 4,284 participants have annual 3MSE scores for up to 6 years of follow-up from baseline. However, there is rapid decline in participant number in subsequent years. The follow-up rates for 11 visit years are provided in Supplementary Table
The WHI Coordinating Center performed imputation against the 1,000 Genomes reference population (Genomes Project et al.,
We focused on the 78 members of the
We applied an additive model and computed the dosage of the major allele from the imputed AA and AB genotype probabilities using the formula: allele dosage = 2 × prob(AA) + prob(AB) (Marchini and Howie,
For cognitive aging, we used two linear regression models. Model 1 was adjusted for baseline age and PC1 to PC3 of population structure. Model 2 was further adjusted for HT group assignment,
We applied two distinct pathway level (i.e., gene-set) methods to calculate the combined effect of
During enrollment to WHIMS, participants were over 65 years old and free of dementia. Demographic and health characteristics of participants at baseline are provided in Table
Demographic characteristics and health and cognitive profiles at baseline.
Age (years) | 70.02 (3.78) |
Depressed mood (CES-D/DIS) |
0.03 (0.1) |
Recreational energy expenditure (MET-hours/week) |
11.67 (13.28) |
Body mass index (kg/m2) | 28.24 (5.52) |
Global cognitive score (3MSE) |
95.92 (3.68) |
<$19,999 | 946 (22%) |
$20K to $34,999 | 1,374 (32%) |
$35K to $49,999 | 920 (22%) |
$50K to $75,999 | 642 (15%) |
>$75K | 365 (9%) |
Don't know or missing | 257 |
<high school | 237 (5%) |
High school or GED | 997 (22%) |
Vocation or some college | 1,812 (40%) |
College graduate | 401 (9%) |
Post-graduate or professional | 1,044 (23%) |
Don't know or missing | 13 |
Estrogen-alone intervention | 712 (16%) |
Estrogen+Progesterone intervention | 1,501 (33%) |
Estrogen-alone control | 747 (17%) |
Estrogen+Progesterone control | 1,544 (34%) |
Never Smoked | 2,344 (52%) |
Past Smoker | 1,808 (41%) |
Current Smoker | 287 (6%) |
Missing | 65 |
Non drinker | 506 (11%) |
Past drinker | 760 (17%) |
<1 drink per month | 551 (12%) |
<1 drink per week | 877 (20%) |
1 to <7 drinks per week | 1,159 (26%) |
7+ drinks per week | 617 (14%) |
Missing | 34 |
No | 2,856 (64%) |
Yes | 1,607 (36%) |
Missing | 41 |
No | 3,658 (82%) |
Yes | 791 (18%) |
Missing | 55 |
No | 3,742 (84%) |
Yes | 707 (16%) |
Missing | 55 |
No | 4,327 (97%) |
Yes | 150 (3%) |
Missing | 27 |
No | 3,370 (75%) |
Yes | 1,134 (25%) |
For survival, the median days from enrollment to end of follow-up or all discovered death is 5,802 (Figure
Survival curve for all-cause mortality. The median time to death is 5,802 days from enrollment (average ±
For cognitive aging, the longitudinal plot of the average 3MSE scores shows marked change over the course of study. As reported previously (Rapp et al.,
Cognitive change over time.
We first tested SNP level association between the 3,693 SNPs/variants and survival time. Model 1 was adjusted for baseline age and population structure, and model 2 was additionally adjusted for
Association plots for aging traits in WHIMS. Each point in the plots represents a variant in an
Quantile-Quantile plots for SNP level association tests.
T/C | 0.88 | 0.04 | 0.21 | 4.9E–28 | |
G/A | 0.84 | 0.01 | 0.21 | 2.0E–23 | |
C/G | 0.96 | 0.42 | −0.05 | 0.0009 | |
G/A | 1.05 | 0.30 | −0.04 | 0.003 | |
C/T | 1.21 | 0.02 | −0.06 | 0.02 | |
C/T | 0.84 | 0.26 | −0.05 | 0.36 |
We used a similar model 1 and model 2 approach for cognitive aging. Two SNPs in LD in
11 | G/C | 0.39 | 9.9E–06 | 41 | 0.0002 | 0.0003 | ||
11 | T/− | 0.12 | 0.0002 | 319 | 0.008 | 0.01 | ||
6 | G/A | 0.16 | 0.0008 | 52 | 0.01 | 0.01 | ||
5 | C/T | 0.42 | 0.0005 | 202 | 0.02 | 0.04 | ||
19 | A/G | 0.03 | 0.02 | 1 | 0.02 | 0.03 | ||
5 | A/G | −0.10 | 0.02 | 9 | 0.05 | 0.06 | ||
3 | A/C | 0.07 | 0.007 | 33 | 0.06 | 0.07 | ||
1 | C/T | 0.03 | 0.04 | 3 | 0.12 | 0.11 | ||
4 | C/T | −0.11 | 0.002 | 262 | 0.13 | 0.15 | ||
6 | G/A | 0.10 | 0.02 | 53 | 0.13 | 0.13 | ||
15 | A/G | −0.05 | 0.02 | 7 | 0.13 | 0.12 | ||
2 | A/G | 0.31 | 0.003 | 289 | 0.15 | 0.15 | ||
3 | G/− | −0.05 | 0.01 | 48 | 0.20 | 0.24 | ||
16 | C/T | −0.04 | 0.02 | 22 | 0.20 | 0.19 | ||
5 | C/T | 0.29 | 0.02 | 37 | 0.21 | 0.22 | ||
6 | G/A | −0.14 | 0.03 | 17 | 0.24 | 0.21 | ||
1 | T/C | −0.04 | 0.02 | 43 | 0.25 | 0.26 | ||
5 | T/C | 0.25 | 0.01 | 73 | 0.26 | 0.25 | ||
8 | A/G | 0.26 | 0.01 | 79 | 0.26 | 0.22 | ||
3 | C/A | 0.20 | 0.02 | 102 | 0.27 | 0.39 | ||
10 | A/T | 0.03 | 0.03 | 23 | 0.27 | 0.34 | ||
7 | G/A | 0.05 | 0.03 | 24 | 0.27 | 0.25 | ||
19 | A/C | −0.11 | 0.03 | 17 | 0.29 | 0.26 | ||
2 | C/− | 0.11 | 0.02 | 31 | 0.31 | 0.30 | ||
9 | T/A | 0.07 | 0.03 | 28 | 0.33 | 0.32 | ||
6 | C/T | 0.06 | 0.04 | 49 | 0.36 | 0.51 | ||
1 | T/C | −0.03 | 0.02 | 64 | 0.40 | 0.34 | ||
17 | A/G | 0.03 | 0.05 | 21 | 0.40 | 0.38 | ||
8 | C/T | −0.09 | 0.02 | 203 | 0.48 | 0.61 | ||
13 | A/G | −0.05 | 0.02 | 96 | 0.50 | 0.39 | ||
12 | −/TTG | 0.07 | 0.02 | 119 | 0.54 | 0.52 | ||
21 | G/T | 0.19 | 0.01 | 148 | 0.57 | 0.48 | ||
1 | rs528474110 | T/− | −0.11 | 0.05 | 27 | 0.62 | 0.48 |
The
We then tested if the
Our main motivation to test the
For the gross outcome, i.e., all-cause mortality, the
Following the analysis of gross survival time, we used cognitive change as a more specific indicator of age-related functional decline, particularly brain aging. Cognitive performance is a strong predictor of health during aging and overall longevity (Riley et al.,
WHIMS was designed to evaluate the impact of HT on cognitive function. Previous studies in WHIMS found that HT resulted in increased risk for cognitive decline (Rapp et al.,
Similar to work on the genetics of age-related cognitive decline (de Jager et al.,
The standard SNP level test treats a single variant as an independent functional unit and fails to capture the summarized effect of multiple variants. Additionally, SNP level associations may have poor replication if the polymorphism is specific to a particular population. For instance, the two significant
A number of different strategies have been developed to test pathway level association. These methods, in essence, treat the gene or gene-set as the functional unit. The underlying statistics can, however, vary greatly from method to method (de Leeuw et al.,
An important limitation is that WHIMS is by no means a representative population. It was specifically designed to study the effect of HT on cognitive function and dementia risk in post-menopausal women. This study benefits from the detailed longitudinal cognitive assessment and demographic and health data. However, any sex-specific effect cannot be accounted for. Additionally, the participants in this study are Caucasians. This limits the generalizability of the association between the
An important next step is to replicate and verify the association of the
To conclude, we provide evidence that variants in
KM: study design, data analysis and interpretation, initial manuscript preparation, and final manuscript approval. BS and SR: helped with data analysis and interpretation, provided access to WHIMS data, and approval of final manuscript. RWW: study design, contributed to manuscript, and final manuscript approval. KJ and RBW: provided oversight and guidance with WHI, contributed to manuscript, and final manuscript approval.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Women's Health Initiative (WHI) program was supported by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health and Human Services (contract numbers HHSN268201100046C, HHSN268201100001C, HHSN268201100002C, HHSN268201100003C, HHSN268201100004C, and HHSN271201100004C). The Women's Health Initiative Memory Study (WHIMS) was funded as an ancillary study to the WHI by Wyeth Pharmaceuticals, Inc., Wake Forest University; and the National Heart, Lung, and Blood Institute, National Institutes of Health; and the National Institute of Aging, National Institutes of Health (contract number HHSN271-2011-00004C). This study was supported by the National Institute of Aging, National Institutes of Health (grant number R01AG043930).
The Supplementary Material for this article can be found online at: