臺灣博碩士論文加值系統

年齡是影響許多表型 (phenotype) 的重要因素。然而，不同族群的老化速度會受基因組成差異與各種環境因素影響而異。為了釐清影響不同族群老化的潛在因素，本研究利用來自臺灣人體生物資料庫 (Taiwan Biobank, TWB) 的132,574人作為研究對象，而其中2,474人有甲基化資料。利用甲基化資料計算Levine等人與Lu等人提出的兩種甲基化年齡。Levine等人提出的甲基化年齡 (DNAm PhenoAge) 是由513個CpG位點組成，Lu等人提出的甲基化年齡 (DNAm GrimAge) 則是由1,030個CpG位點組成。
DNAm PhenoAge係利用由實際年齡與其他9個臨床標記 (clinical marker) 構成之表觀年齡 (phenotypic age) 以國家健康營養問卷第三版 (National Health and Nutrition Examination Survey III, NHANES III) 作為訓練集推導而來，而DNAm GrimAge則是由性別、真實年齡與其它8個甲基化標記 (methylation-based marker) 估算而成。我們進一步檢驗這些標記與實際年齡的相關性，發現所有標記與真實年齡的相關性均可發現於TWB。然而，DNAm PhenoAge和DNAm GrimAge對於臺灣人生理年齡的預測能力與真實年齡並無太大差異。
在2,474人中有2,392人 (約占97%) 的DNAm PhenoAge比實際年齡低。我們進一步以NHANES III推導由6個臨床標記計算而成的表觀年齡 (6-marker phenotypic age)，發現其與DNAm PhenoAge高度相關且在132,574人中有128,471人 (約占97%) 的6-marker phenotypic age比實際年齡低。然而，2,474人中僅1,084人 (約占44%) 的DNAm GrimAge 比實際年齡低，且其平均值比DNAm PhenoAge的平均值高約10歲。為了釐清此差異的原因，本研究比較TWB中可取得的臨床標記與NHANES III的組成差異。我們發現所有TWB可取得的臨床標記其於血液中的濃度與NHANES III相比，無論男女皆有顯著差異，此現象使得TWB的受試者DNAm PhenoAge較為年輕。
因為Levine等人提出的甲基化年齡是由美國人估算而成，故此方法可能會因種族因素產生偏差。本研究進一步分析513個CpG位點並探究其與實際年齡的關係。在調整了性別、身體質量指數、運動習慣、飲酒習慣、抽菸習慣、教育水準與批次效應 (batch effect) 後，我們發現在513個位點中226個位點與實際年齡相關且當中的224個位點其與實際年齡的相關性與Levine等人的研究一致。
總結而言，本研究發現以美國人為主體的研究中約99% (224/226) 與實際年齡相關的CpG位點可以被臺灣族群驗證。另外，本研究發現真實年齡與兩種甲基化年齡所使用的臨床標記之間的關係也可驗證於臺灣人，且他們與真實年齡的相關性皆很高。考量TWB中的DNAm PhenoAge傾向於低估，我們認為DNAm GrimAge較適用於臺灣人。

Age is considered one of the most crucial covariates that affect phenotypes. However, aging rate may vary among different populations due to genetic variation or miscellaneous environmental exposures. To clarify the underlying causes, this study utilized data from 132,574 subjects provided by Taiwan biobank (TWB), wherein DNA methylation data were available for 2,474 of the subjects. We further computed DNA methylation age using the method proposed by Levine et al. (DNAm PhenoAge) utilizing the 513 methylation sites and the method proposed by Lu et al. (DNAm GrimAge) utilizing 1,030 sites.
DNAm PhenoAge was derived from phenotypic age comprising chronological age and 9 clinical markers using National Health and Nutrition Examination Survey III (NHANES III) as training data, while DNAm GrimAge was derived from sex, chronological age and 8 methylation-based markers. We further examined the correlations between chronological age and the markers used for DNAm PhenoAge and DNAm GrimAge and discovered that all of them can be found in TWB. However, we did not find much difference between DNAm GrimAge, DNAm PhenoAge and chronological age in terms of the predictability of biological age in Taiwanese people.
Among the 2,474 subjects, 2,392 (97%) were younger in terms of DNAm PhenoAge as compared to chronological age. We further derived a 6-marker phenotypic age computed from the 6 clinical markers available in TWB and found that it was highly correlated with DNAm PhenoAge. We also found that among the 132,574 subjects, 128,471 (97%) were younger in terms of 6-marker phenotypic age as compared to chronological age. However, among the 2,474 subjects, only 1,084 (44%) were younger in terms of DNAm GrimAge as compared to chronological age, and the mean value of DNAm GrimAge was about 10 years higher as compared to that of DNAm PhenoAge. To clarify the possible causes for the discrepancy, our study compared the difference in levels of the available markers among TWB and NHANES III. We found the levels of all the available clinical markers in TWB to be significantly different as compared to those of available markers in NHANES III regardless of gender, which could cause the DNAm PhenoAge for subjects in TWB to be younger.
Since Levine’s methylation age was derived from Americans and can be racially biased, this study further conducted correlation analyses on the 513 methylation sites and examined their relationships with age. After adjusting for sex, body mass index, physical activity, alcohol drinking status, cigarette smoking status, educational attainment and batch effects, 226 out of the 513 methylation sites were identified to be associated with chronological age and 224 of them are associated in the same correlation direction of Levine et al.’s study.
In summary, our study shows that ~99% (224/226) of age associated CpG sites discovered in Americans have the same associations with age in Taiwanese people. In addition, our study shows that the relationship between chronological age and the clinical markers used in DNAm PhenoAge and the relationship between chronological age and the clinical markers used in DNAm GrimAge can also be verified in Taiwanese people. Moreover, the correlations between chronological age and the two methylation ages are both high. Considering the tendency for underestimation of DNAm PhenoAge in TWB, we conclude that DNAm GrimAge is a more appropriate approach for applications in Taiwanese people.

口試委員審定書 i
謝辭 ii
中文摘要 iii
Abstract v
Contents viii
List of Figures x
List of Tables xi
1.Introduction 1
2.Literature review 3
3.Materials and Methods 8
3.1 Study design 8
3.2 TWB 8
3.3 NHANES III 10
3.4 Methylation age calculation 11
3.5 Phenotypic age calculation 11
3.6 The mixed effects model for identifying age-associated sites 12
3.7 Agreement analyses 13
3.8 Correlation analyses 13
4.Results 14
4.1 Basic characteristics of TWB and NHANES III 14
4.2 DNAm PhenoAge, DNAm GrimAge and chronological age in TWB 14
4.3 Phenotypic age among TWB and NHANES III 14
4.4 Agreement analyses for phenotypic age, DNAm PhenoAge and DNAm GrimAge 15
4.5 Correlation analyses between chronological age, the 5 clinical markers and the 8 methylation-based markers 15
4.6 The mean levels of the 6 available markers among TWB and NHANES III 16
4.7 Methylation levels and environmental exposures 16
5.Conclusions 19
6.Discussion 22
6.1 Possible explanations 22
6.2 Limitations 22
7.Figures 26
8.Tables 30
9.Reference 33
10.Supplementary 36
Table S1 The regression coefficients derived from formula (5) and the Pearson’s correlation coefficients between chronological age and the methylation levels of the 513 sites used in DNAm PhenoAge among TWB and Levine’s study 36
Table S2 The Pearson’s correlation coefficients between chronological age and the 8 methylation-based biomarkers in TWB 63
Table S3 The converting table from level to year(s) for educational attainment in TWB 63

Crimmins, E., Vasunilashorn, S., Kim, J. K., & Alley, D. (2008). Biomarkers related to aging in human populations. Advances in clinical chemistry, 46, 161-216.
Crimmins, E. M., Thyagarajan, B., Levine, M. E., Weir, D. R., & Faul, J. (2021). Associations of Age, Sex, Race/Ethnicity, and Education With 13 Epigenetic Clocks in a Nationally Representative US Sample: The Health and Retirement Study. The Journals of Gerontology: Series A.
Cronjé, H. T., Nienaber-Rousseau, C., Min, J. L., Green, F. R., Elliott, H. R., & Pieters, M. (2021). Comparison of DNA methylation clocks in black South African men. Epigenomics, 13(06), 437-449.
Di Lena, P., Sala, C., Prodi, A., & Nardini, C. (2019). Missing value estimation methods for DNA methylation data. Bioinformatics, 35(19), 3786-3793.
Di Lena, P., Sala, C., Prodi, A., & Nardini, C. (2020). Methylation data imputation performances under different representations and missingness patterns. BMC bioinformatics, 21(1), 1-22.
Dolcini, J., Wu, H., Nwanaji-Enwerem, J. C., Kiomourtozlogu, M.-A., Cayir, A., Sanchez-Guerra, M., . . . Baccarelli, A. A. (2020). Mitochondria and aging in older individuals: an analysis of DNA methylation age metrics, leukocyte telomere length, and mitochondrial DNA copy number in the VA normative aging study. Aging (Albany NY), 12(3), 2070.
Fortin, J.-P., Triche Jr, T. J., & Hansen, K. D. (2017). Preprocessing, normalization and integration of the Illumina HumanMethylationEPIC array with minfi. Bioinformatics, 33(4), 558-560.
Hannum, G., Guinney, J., Zhao, L., Zhang, L., Hughes, G., Sadda, S., . . . Gao, Y. (2013). Genome-wide methylation profiles reveal quantitative views of human aging rates. Molecular cell, 49(2), 359-367.
Hill, N. R., Fatoba, S. T., Oke, J. L., Hirst, J. A., O’Callaghan, C. A., Lasserson, D. S., & Hobbs, F. R. (2016). Global prevalence of chronic kidney disease–a systematic review and meta-analysis. PloS one, 11(7), e0158765.
Hillary, R. F., Stevenson, A. J., Cox, S. R., McCartney, D. L., Harris, S. E., Seeboth, A., . . . Redmond, P. (2019). An epigenetic predictor of death captures multi-modal measures of brain health. Molecular psychiatry, 1-11.
Horvath, S. (2013). DNA methylation age of human tissues and cell types. Genome biology, 14(10), 1-20.
Houseman, E. A., Accomando, W. P., Koestler, D. C., Christensen, B. C., Marsit, C. J., Nelson, H. H., . . . Kelsey, K. T. (2012). DNA methylation arrays as surrogate measures of cell mixture distribution. BMC bioinformatics, 13(1), 1-16.
Jones, M. J., Goodman, S. J., & Kobor, M. S. (2015). DNA methylation and healthy human aging. Aging cell, 14(6), 924-932.
Lau, L. H., Lew, J., Borschmann, K., Thijs, V., & Ekinci, E. I. (2019). Prevalence of diabetes and its effects on stroke outcomes: A meta‐analysis and literature review. Journal of diabetes investigation, 10(3), 780-792.
Levine, M. E., & Crimmins, E. M. (2018). Is 60 the new 50? Examining changes in biological age over the past two decades. Demography, 55(2), 387-402.
Levine, M. E., Lu, A. T., Quach, A., Chen, B. H., Assimes, T. L., Bandinelli, S., . . . Li, Y. (2018). An epigenetic biomarker of aging for lifespan and healthspan. Aging (Albany NY), 10(4), 573.
Lu, A. T., Quach, A., Wilson, J. G., Reiner, A. P., Aviv, A., Raj, K., . . . Stewart, J. D. (2019). DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging (Albany NY), 11(2), 303.
McCrory, C., Fiorito, G., Hernandez, B., Polidoro, S., O’Halloran, A. M., Hever, A., . . . Vineis, P. (2021). GrimAge outperforms other epigenetic clocks in the prediction of age-related clinical phenotypes and all-cause mortality. The Journals of Gerontology: Series A, 76(5), 741-749.
Philibert, R., Beach, S. R., Lei, M.-K., Gibbons, F. X., Gerrard, M., Simons, R. L., & Dogan, M. V. (2020). Array-based epigenetic aging indices may be racially biased. Genes, 11(6), 685.
Sebastiani, P., Thyagarajan, B., Sun, F., Schupf, N., Newman, A. B., Montano, M., & Perls, T. T. (2017). Biomarker signatures of aging. Aging cell, 16(2), 329-338.
Tajuddin, S. M., Hernandez, D. G., Chen, B. H., Hooten, N. N., Mode, N. A., Nalls, M. A., . . . Zonderman, A. B. (2019). Novel age-associated DNA methylation changes and epigenetic age acceleration in middle-aged African Americans and whites. Clinical epigenetics, 11(1), 1-16.
Triche Jr, T. J., Weisenberger, D. J., Van Den Berg, D., Laird, P. W., & Siegmund, K. D. (2013). Low-level processing of Illumina Infinium DNA methylation beadarrays. Nucleic acids research, 41(7), e90-e90.
Wang, C., Koutrakis, P., Gao, X., Baccarelli, A., & Schwartz, J. (2020). Associations of annual ambient PM2. 5 components with DNAm PhenoAge acceleration in elderly men: the Normative Aging Study. Environmental Pollution, 258, 113690.
Zhao, W., Ammous, F., Ratliff, S., Liu, J., Yu, M., Mosley, T. H., . . . Smith, J. A. (2019). Education and lifestyle factors are associated with DNA methylation clocks in older African Americans. International journal of environmental research and public health, 16(17), 3141.
Zou, J., Lippert, C., Heckerman, D., Aryee, M., & Listgarten, J. (2014). Epigenome-wide association studies without the need for cell-type composition. Nature methods, 11(3), 309-311.