ALDH2 p.E504K Variation and Sex Are Major Factors Associated with Current and Quitting Alcohol Drinking in Japanese Oldest Old

This study identified the factors associated with current and quitting alcohol drinking in the Oldest Old to better understand the associated factors and mechanisms underlying drinking behaviors in this age group. Results of a questionnaire for drinking behavior in 1015 Japanese Oldest Old citizens aged 85 to 89 years revealed that 56.0% of men and 24.0% of women were current drinkers. A genome-wide association study revealed that the rs671 G > A variation, which corresponds to the aldehyde dehydrogenase 2 (ALDH2) p.E504K missense variant, was significantly associated with current drinking (odds ratio: 3.8, p = 3.33 × 10−31). Variable selection with 41 factors and multivariate regression logistic analysis for current drinking indicated that the rs671 genotype and sex were the most significant factors in the Oldest Old. Further analysis revealed that the rs671 genotype, alcohol-associated biomarkers, a history of heart or kidney disease, and frailty score are factors associated with quitting drinking in the Oldest Old men, whereas smoking history, walking time, and depression score were factors associated with quitting drinking in the Oldest Old women. These results indicate that the ALDH2 p.E504K variation is a major factor associated with current and quitting drinking in the Japanese Oldest Old.


Introduction
Alcohol drinking (drinking) and its effects have been present in societies throughout human history. Social drinking with low to moderate alcohol consumption is thought to be a pleasurable behavior; however, habitual drinking with excess alcohol consumption results in several negative outcomes; it is a risk factor for poor health, road incidents, and alcohol dependence. A global status report on alcohol and health by the World Health Organization (WHO) revealed that 57% (3.1 billion people) of the global population aged 15 years and over had abstained from drinking in the previous 12 months, and 2.3 billion people were current drinkers in 2016 [1]. Alcohol is consumed by more than half of the population in only three world regions, the Americas, Europe, and the Western Pacific, indicating that drinking is biased across regions and civilizations worldwide. In Japan, 76.4% of men and 61.9% of women aged 15 years and over consumed alcohol in 2009, ranking 22nd for men and 30th for women among drinking countries and highest in East Asia [2].
The age of drinking is associated with an important health issue. Globally, 26.5% of adolescents aged 15-19 years are current drinkers, accounting for 155 million adoles-

Study Population
This study used data from cohort studies of the KAWP [15]. The inclusion criteria of the KAWP were as follows: (1) individuals aged between 85-89 and being a resident of Kawasaki, a city with a population of 1.5 million in the Kanagawa prefecture; (2) individuals having no limitations in the basic activities of daily living; and (3) individuals being able to visit the hospital for cohort survey by themselves. A total of 1026 independent older adults were enrolled in the KAWP and a comprehensive baseline assessment was done, including questionnaires for drinking behavior and life status, cognitive and mental status, disease histories, measurement of physical function, and plasma biomarkers ( Table 1). Among these participants, 1015 Oldest Old persons (512 men and 503 women) who responded to the questionnaire regarding drinking behavior and accepted genome DNA sequence analysis were included in this study. The KAWP was approved by the ethics committee of the Keio University School of Medicine (ID: 20160297) and was registered in the University Hospital Medical Information Network Clinical Trial Registry as an observational study (ID: UMIN000026053).

Baseline Characteristics
Methods for collecting baseline characteristics of KAWP including the questionnaire for life status and disease histories, measurement of cognitive and mental statuses, measurement of physical functions, and measurement of biomarkers in plasma have been described previously [15]. Briefly, the questionnaire and health data were obtained from the KAWP baseline survey conducted between March 2017 and December 2018. All participants were invited to hospitals, interviewed, and examined using a study protocol that was designed by the Tokyo Oldest Old Survey on Total Health and Japan Semi-supercentenarian Study, which were previously managed by the Center for Supercentenarian Medical Research, Keio University School of Medicine [17,18].

Genotyping
Total genomic DNA was extracted from whole blood using a FlexGene DNA Kit (QIAGEN, Hilden, Germany). The genotypes of 0.65M SNVs were determined using the Infinium Asian Screening Array-24 v1.0 BeadChip kit, according to the manufacturer's standard protocol. All microarray scan images were analyzed and then the SNV genotypes were determined using GenomeStudio (version 2013, Illumina, San Diego, CA, USA). Quality filters for both samples and SNV genotypes were also applied by GenomeStudio. No sample was filtered using an SNV call rate filter (cutoff < 98.0, lowest call rate: 0.9803). In addition, 15,781 out of 657,060 SNVs (2.4%) were filtered out by SNV genotyping quality control filters (R Mean (AA, AB, BB) < 0.2500 and Cluster Seq < 0.4000). Finally, the genotypes of 641,279 SNVs from 1016 KAWP participants were obtained.

Pre-Phasing and Imputation
Genotype data on the Infinium Asian Screening Array were converted to VCF format by GenomeStudio (version 2013). These genotype data were prephased by shapeit4.0 program according to documentation with default parameters and genetic map sets (ge-netic_maps.b37) on the shapit4 web site (https://odelaneau.github.io/shapeit4/, accessed on 5 February 2021) [19]. The SNVs were imputed using prephased genotype data using the impute5 program (version1.1.4, https://jmarchini.org/impute5/, accessed on 5 February 2021) with default parameters and the phased VCF data on the 1000Genomes phase 3 data (ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/release/20130502/, accessed on 5 February 2021) [20]. To evaluate the probability of accuracy for imputed variants, "info" metric values calculated by the impute5 program were employed against each imputed variant, and 0.5 was used as a threshold for the impute5 info value as recommended on the web site. Among 80.9 million SNVs imputed by imputed5, 6.38 million SNVs were screened using combination filters for the imputation reliability filter (impute5 info ≥ 0.5), minor allele frequency ≥ 0.02, and p-value for Hardy-Weinberg equilibrium >10 −6 as high-quality SNV data.

Genome-Wide Association Study
To identify SNVs associated with drinking behavior in the Oldest Old, this study analyzed the genome-wide association between the answers for the drinking behavior questionnaire. GWAS was performed using linear models with sex information as covariate by the PLINK program (version 1.90) (version 1.90) [21]. The genomic inflation est. lambda (based on median chisq) was 1.014, indicating that no evident inflation occurred.

Statistical Analysis
The difference in baseline data was evaluated using the t-test, Wilcoxon rank-sum test, or chi-square test (Table 1). LASSO is a penalized technique for variable selection that is effective when the number of events per variable is low [16]. LASSO shrinks the coefficients for weaker predictors to zero. The degree of shrinkage was determined by the optimal parameter lambda, as identified by five-fold cross-validation. Multivariate logistic regression analyses were performed using the generalized linear model using factors selected by LASSO. All statistical analyses were performed using the R script (version 4.0.3) with effsize (cohen.d function for effect size estimation of t-test data (version 0.8.1)), vcd (assocstats function for effect size estimation of Chi-square test data (version 1.4-8), stat (glm function for multivariate regression analysis (version 4.0.3)), glmnet (glmnet function for LASSO (version 4.1)), exactRankTests (wilcox.exact function for Wilcoxon rank-sum test (version 0.8-31)), rstatix (wilcox_effsize function for effect size estimation of wilcoxon rank-test data (version 0.7.0)) and default packages.

Baseline Characteristics of the Kawasaki Aging and Wellbeing Project Cohort
A baseline survey was conducted with the participants of the KAWP, an ongoing longitudinal cohort study of older adults aged 85 years or older [15]. The participants' characteristics are presented in Table 1. The total number of participants with genomic data analyzed through DNA microarray was 1015, of which 287 men (56.0%) and 121 women (24.0%) were current drinkers (Figure 1a).

Baseline Characteristics of the Kawasaki Aging and Wellbeing Project Cohort
A baseline survey was conducted with the participants of the KAWP, an ongoing longitudinal cohort study of older adults aged 85 years or older [15]. The participants' characteristics are presented in Table 1. The total number of participants with genomic data analyzed through DNA microarray was 1015, of which 287 men (56.0%) and 121 women (24.0%) were current drinkers (Figure 1a).

Imputation and GWAS for Current Drinking in Japanese Oldest Old
To detect genetic variants associated with current drinking in the Oldest Old, this study analyzed SNVs by a GWAS with a two-choice questionnaire (yes/no) for current drinking, using the imputed DNA sequences of Japanese Oldest Old (Figure 1b). The 0.65 million SNVs were genotyped using the Infinium Asian Screening Array against the 1016 Oldest Old. These SNV data were prephased and then imputed using 1000Genomes phase

Imputation and GWAS for Current Drinking in Japanese Oldest Old
To detect genetic variants associated with current drinking in the Oldest Old, this study analyzed SNVs by a GWAS with a two-choice questionnaire (yes/no) for current drinking, using the imputed DNA sequences of Japanese Oldest Old (Figure 1b). The 0.65 million SNVs were genotyped using the Infinium Asian Screening Array against the 1016 Oldest Old. These SNV data were prephased and then imputed using 1000Genomes phase 3 variation data. After filtering using the imputation reliability, allele frequency, and Hardy-Weinberg equilibrium filters, the genotypes of 6.38 million imputed SNVs were obtained. We then analyzed the association between the two-choice questionnaire for current drinking and genotypes in 1015 older adults using the plink 1.90 program. As a result, rs671 G > A SNV, which corresponds to the ALDH2 p.E504K missense variant, was detected as the most significant variant (odds ratio: 3.8, p = 3.33 × 10 −31 , adjusted P (Bonferroni) = 2.17 × 10 −24 ) associated with current drinking in the Japanese Oldest Old.

ALDH2 p.E504K Variation Is Associated with Drinking Behavior, Frequency, and Amount in the Japanese Oldest Old
To confirm the association between current drinking and ALDH2 p.E504K variation in the Oldest Old, the frequency of current drinkers was compared in each genotype (Figure 2a). The results revealed that 82.8% of 250 Oldest Old men without the ALDH2 p.E504K variation were current drinkers; however, 36.4% of 220 and 0% of 42 Oldest Old men with the heterozygous and homozygous ALDH2 p.E504K variation were current drinkers, respectively. In women, 33.7% of 288 Oldest Old women without the ALDH2 p.E504K variation were current drinkers; however, 12.1% of 190 and 4.0% of 25 Oldest Old women with the heterozygous and homozygous ALDH2 p.E504K were current drinkers, respectively. These results indicate that the ALDH2 p.E504K variation is significantly associated with current drinkers among Oldest Old men and women.
p.E504K variation were current drinkers; however, 36.4% of 220 and 0% of 42 Oldest Old men with the heterozygous and homozygous ALDH2 p.E504K variation were current drinkers, respectively. In women, 33.7% of 288 Oldest Old women without the ALDH2 p.E504K variation were current drinkers; however, 12.1% of 190 and 4.0% of 25 Oldest Old women with the heterozygous and homozygous ALDH2 p.E504K were current drinkers, respectively. These results indicate that the ALDH2 p.E504K variation is significantly associated with current drinkers among Oldest Old men and women.  (c) Distribution of heavy and light drinkers among frequent drinkers by genotype of ALDH2 p.E504K variation. Black: heavy drinker (drinking ≥ 633 mL beer/day); dark gray: light drinker (drinking < 633 mL beer/day). (d) Distribution of heavy and light drinkers among non-frequent drinkers by genotype of ALDH2 p.E504K variation. Light gray: heavy drinker (drinking ≥ 633 mL beer/day); dark gray: light drinker (drinking < 633 mL beer/day). Furthermore, this study analyzed the association between drinking frequency and ALDH2 p.E504K variation in current drinkers (Figure 2b). An individual who drinks three times or more per week was defined as a "frequent drinker." As a result, 83.1% of 207 current drinkers among the Oldest Old men without the ALDH2 p.E504K variation is considered as significantly frequent compared to the 63.3% of 79 current drinkers among the Oldest Old men with the ALDH2 p.E504K variation. This study found a frequency difference between current drinkers among Oldest Old women without ALDH2 p.E504K variation and those with ALDH2 p.E504K variation; however, the difference was not significant.
Finally, the association between the drinking amount and ALDH2 p.E504K variation in frequent and non-frequent drinkers was analyzed in this study (Figure 2c,d). An individual who drinks over 633 mL of beer and less or equal to 633 mL of beer per day was defined as a "heavy drinker" and "light drinker," respectively. As a result, 33.1% of 172 frequent drinkers among the Oldest Old men without the ALDH2 p.E504K variation were heavy drinkers and were significantly more frequent than the 12.0% of 50 frequent drinkers among the Oldest Old men with ALDH2 p.E504K variation. For other combinations, including non-frequent drinkers among the Oldest Old men and both frequent and non-frequent drinkers among the Oldest Old women, no significant difference in alcohol consumption was observed.

Distribution of Age at Quitting Drinking in Oldest Old Men and Women with the ALDH2 p.E504K Variation
To clarify the associations among frequency of quitting drinking, ALDH2 p.E504K variation, and age at quitting drinking in the Oldest Old, frequency and age distribution of past drinkers were analyzed. In the Oldest Old men, the frequency of past drinkers was significantly different depending on the ALDH2 p.E504K variation (Figure 3a). Distribution of age at quitting drinking in the Oldest Old men indicated that the frequency of past drinkers without ALDH2 p.E504K variation mainly started to increase at the age of 60 and continued to increase until the age of 80; however, the frequency of past drinkers with heterozygous ALDH2 p.E504K variation started to increase in their 50s and peaked in their 60s, indicating that the ALDH2 p.E504K variation would also affect the age at quitting drinking in Japanese men (Figure 3b). For women, no significant difference was found between the frequency of past drinkers by ALDH2 p.E504K variation (Figure 3c). The distribution of age at quitting drinking in the Oldest Old women indicated that past drinkers without the ALDH2 p.E504K variation in women showed a pattern similar to that in men; however, no evident pattern was observed for past drinkers with heterozygous ALDH2 p.E504K variation in women because there were only eight past drinkers with heterozygous ALDH2 p.E504K variation (Figure 3d). These results suggest that ALDH2 p.E504K variation does not affect the age at quitting drinking in Japanese women.

Variable Selection Associated with Current Drinking by LASSO and Its Multivariate Regression Logistic Analysis
For multivariate regression logistic analysis associated with current drinking, 41 factors were selected, including two known drinking behavior-associated genetic factors (genotypes of rs671 in ALDH2 and rs1229984 in ADH1B) and 39 factors from the baseline survey of the KAWP (sex, questionnaire for drinking behavior and life status, cognitive and mental statuses, disease histories, measurement of physical function, and biomarkers in plasma) of the Oldest Old (Table 1). To select the variables for multivariate regression logistic analysis, the LASSO method was employed with five-fold cross-validation against 412 Oldest Old men and 385 Oldest Old women without missing values. As a result of five-fold cross-validation, 23 factors were selected for further multivariate logistic analysis

Variable Selection Associated with Current Drinking by LASSO and Its Multivariate Regression Logistic Analysis
For multivariate regression logistic analysis associated with current drinking, 41 factors were selected, including two known drinking behavior-associated genetic factors (genotypes of rs671 in ALDH2 and rs1229984 in ADH1B) and 39 factors from the baseline survey of the KAWP (sex, questionnaire for drinking behavior and life status, cognitive and mental statuses, disease histories, measurement of physical function, and biomarkers in plasma) of the Oldest Old (Table 1). To select the variables for multivariate regression logistic analysis, the LASSO method was employed with five-fold cross-validation against 412 Oldest Old men and 385 Oldest Old women without missing values. As a result of five-fold cross-validation, 23 factors were selected for further multivariate logistic analysis for the Oldest Old (Figure 4a). As a result of multivariate regression logistic analysis with these 23 factors, 8 significant factors, including 4 positively associated factors (Body Mass Index (BMI), high-density lipoprotein cholesterol (HDLC), years of education, and γ glutamyl transpeptidase (γGTP) and 4 negatively associated factors (genotype of rs671 (ALDH2), sex (women), alanine transaminase (ALT), and history of renal disease) were found (Figure 4b).
We also analyzed men and women separately, and 13 and 5 factors were selected for further multivariate logistic analysis for Oldest Old men and women, respectively (Supplementary Figures S1a and S2a). As a result of multivariate regression logistic analysis with these 13 factors for Oldest Old men, nine significant factors, including six positively associated factors (HDLC, smoking history, cholinesterase (CHE), years of education, γ GTP, and BMI) and three negatively associated factors (genotype of rs671 (ALDH2), alkaline phosphatase (ALP), and history of heart disease), were found (Supplementary Figure S1b). Furthermore, only one of the five factors examined, rs671 (ALDH2), was significant and negatively associated with current drinking in Oldest Old women (Supplementary Figure S2b).

Variable Selection Associated with Quitting Drinking by LASSO and Its Multivariate Regression Logistic Analysis
For multivariate regression logistic analysis associated with quitting drinking, this study selected 230 current and 81 past drinkers among Oldest Old men and 91 current and 36 past drinkers among Oldest Old women. Moreover, 41 factors were selected, including two known drinking behavior-associated genetic factors (genotypes of rs671 in ALDH2 and rs1229984 in ADH1B) and 39 factors from the baseline survey of the KAWP. As a result of LASSO with five-fold cross-validation, 15 and 4 factors were selected for further multivariate logistic analysis for Oldest Old men and women, respectively ( Figure  5a,b). As a result of multivariate regression logistic analysis for Oldest Old men, this study found nine significant factors, including six positively associated factors (genotype of

Variable Selection Associated with Quitting Drinking by LASSO and Its Multivariate Regression Logistic Analysis
For multivariate regression logistic analysis associated with quitting drinking, this study selected 230 current and 81 past drinkers among Oldest Old men and 91 current and 36 past drinkers among Oldest Old women. Moreover, 41 factors were selected, including two known drinking behavior-associated genetic factors (genotypes of rs671 in ALDH2 and rs1229984 in ADH1B) and 39 factors from the baseline survey of the KAWP. As a result of LASSO with five-fold cross-validation, 15 and 4 factors were selected for further multivariate logistic analysis for Oldest Old men and women, respectively (Figure 5a,b). As a result of multivariate regression logistic analysis for Oldest Old men, this study found nine significant factors, including six positively associated factors (genotype of rs671 (ALDH2), ALT, ALP, history of heart disease, frailty score, and history of renal disease) and three negatively associated factors (γGTP, CHE, and HDLC; Figure 5a). Additionally, only one factor (past smoking) was found to be a significantly positively associated with these four factors for Oldest Old women, through multivariate regression logistic analysis (Figure 5b).

Discussion
In this study, we identified the factors associated with both current and quitting drinking against 1015 Japanese aged 85 to 89 years from the KAWP cohort. The questionnaire for drinking behavior against the KAWP cohort revealed that 56.0% of men and 24.0% of women were current drinkers, which is comparable to a previous report on Japanese drinking behavior [6]. A GWAS revealed that the rs671 SNV, which corresponds to the ALDH2 p.E504K missense variant, is significantly associated with current drinking. Further analysis revealed that the ALDH2 p.E504K variation is also associated with quitting drinking in Oldest Old men; however, only past smoking was associated with quitting drinking in Oldest Old women. The variable selection by LASSO and further multivariate regression logistic analysis indicated that the genotype of rs671 and sex are significantly associated with current drinking among a comprehensive set of 41 investigated factors in the Oldest Old. These results indicate that the ALDH2 p.E504K variation is one of the major factors for several drinking behaviors, including current and quitting drinking, even in the Japanese Oldest Old.
Previous GWAS results revealed that both ADH1B and ALDH2 variants are associated with alcohol dependency in Chinese people; however, only the ALDH2 variant is associated with drinking behavior in Japanese people [12,14]. In this study, GWAS for current drinking revealed that the ALDH2 p.E504K variation was associated with current drinking even in the Japanese Oldest Old; however, no association was found between current drinking and ADH1B variation, which is consistent with previous studies [12]. These results suggest that the ALDH2 p.E504K variation negatively affects drinking behaviors, including current and quitting drinking, by slowing the clearance of acetaldehyde from the blood.
The results of this study also revealed that the frequency of quitting drinking was

Discussion
In this study, we identified the factors associated with both current and quitting drinking against 1015 Japanese aged 85 to 89 years from the KAWP cohort. The questionnaire for drinking behavior against the KAWP cohort revealed that 56.0% of men and 24.0% of women were current drinkers, which is comparable to a previous report on Japanese drinking behavior [6]. A GWAS revealed that the rs671 SNV, which corresponds to the ALDH2 p.E504K missense variant, is significantly associated with current drinking. Further analysis revealed that the ALDH2 p.E504K variation is also associated with quitting drinking in Oldest Old men; however, only past smoking was associated with quitting drinking in Oldest Old women. The variable selection by LASSO and further multivariate regression logistic analysis indicated that the genotype of rs671 and sex are significantly associated with current drinking among a comprehensive set of 41 investigated factors in the Oldest Old. These results indicate that the ALDH2 p.E504K variation is one of the major factors for several drinking behaviors, including current and quitting drinking, even in the Japanese Oldest Old.
Previous GWAS results revealed that both ADH1B and ALDH2 variants are associated with alcohol dependency in Chinese people; however, only the ALDH2 variant is associated with drinking behavior in Japanese people [12,14]. In this study, GWAS for current drinking revealed that the ALDH2 p.E504K variation was associated with current drinking even in the Japanese Oldest Old; however, no association was found between current drinking and ADH1B variation, which is consistent with previous studies [12]. These results suggest that the ALDH2 p.E504K variation negatively affects drinking behaviors, including current and quitting drinking, by slowing the clearance of acetaldehyde from the blood.
The results of this study also revealed that the frequency of quitting drinking was significantly associated with ALDH2 p.E504K variation in the Oldest Old men. Multivariate regression logistic analysis between current and past drinkers among the Oldest Old men revealed that ALDH2 p.E504K variation, biomarkers associated with drinking, disease histories for heart and kidney, and frailty score were associated with quitting drinking. Among these factors, the frailty score and disease histories for the heart and kidney suggest that Older men with health problems tend to quit drinking. Furthermore, the average eGFR calculated with creatinine in past drinkers without ALDH2 p.E504K variation (homozygotes of rs761 major allele) in their 70-80s is significantly lower than that in the early quit or current drinkers without ALDH2 p.E504K variation. These results suggest that health deterioration could be one of the reasons for the increasing number of past drinkers without the ALDH2 p.E504K variation after their 60s, and appropriate health guidance based on ALDH2 p.E504K genotype by health care providers could be effective to prevent alcohol related-health problems for the Oldest Old. This study also revealed that the frequency of past drinkers with heterozygous ALDH2 p.E504K variation started to increase in their 50s and peaked in their 60s. Although there is no confirmatory data to explain the reason for this peak, it was assumed that most Japanese men retired from their jobs in their 60s, and the social behavior changes associated with job retirement could cause a decrease in the chance of drinking and promote quitting drinking in Japan. Another possibility is that a low capability for alcohol/aldehyde degradation by ALDH2 p.E504K variation caused low alcohol tolerance, resulting in it being easier to quit drinking.
In this study, the cohort data analysis with the DNA microarray data enable us to calculate multivariate regression logistic analysis with both environmental and genetic factors. The combination of cohort and genetic data makes it possible to analyze the gene and environmental interaction. Although only two genetic factors are strongly associated with current drinking, our data revealed that the KAWP dataset is valuable not only as an Oldest Old cohort, but also cohort data in which both environmental and genetic factors are available.
This study has several limitations. First, the size of our cohort was small for GWAS; therefore, our GWAS could only detect strongly-associated genetic factors. Furthermore, it has already been reported that the ALDH2 p.E504K variation is associated with the risk of esophageal and stomach cancers in East Asia [22]. This study could not find a significant association between ALDH2 p.E504K variation and esophageal cancer history because only four cases of esophageal cancer were observed in the entire KAWP cohort, indicating that a larger cohort is needed to verify the association between alcohol-related cancers and ALDH2 p.E504K variation. Second, we could not measure all factors associated with drinking behaviors although we measured a comprehensive set of factors potentially associated with drinking behaviors. Another limitation is that the KAWP cohort participants included relatively healthy older adults; therefore, older adults who have alcohol disorders, dementia, and depression might have been possibly less frequent. In future studies, this cohort will be developed to clarify the association between genotype and these diseases.

Conclusions
This study analyzed the factors associated with the drinking behavior of the Oldest Old in Japan by GWAS and multivariate regression logistic analysis. It was revealed that ALDH2 p.E504K variation and sex were the major factors associated with current and quitting drinking in the Japanese Oldest Old. This study also suggested that age at quitting drinking was associated with ALDH2 p.E504K variation although other environmental factors might also be associated in the Oldest Old.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/genes12060799/s1, Figure S1: Variable selection associated with current drinking by LASSO and its multivariate regression logistic analysis against Oldest Old men, Figure S2: Variable selection associated with current drinking by LASSO and its multivariate regression logistic analysis against Oldest Old women. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.