Abstract
Background
DNA analysis for forensic investigations has a long tradition with important developments and optimizations since its first application. Traditionally, short tandem repeats analysis has been the most powerful method for the identification of individuals. However, in addition, epigenetic changes, i.e., DNA methylation, came into focus of forensic DNA research. Chronological age prediction is one promising application to allow for narrowing the pool of possible individuals who caused a trace, as well as to support the identification of unknown bodies and for age verification of living individuals.
Objective
This review aims to provide an overview of the current knowledge, possibilities, and (current) limitations about DNA methylation-based chronological age prediction with emphasis on forensic application.
Methods
The development, implementation and application of age prediction tools requires a deep understanding about the biological background, the analysis methods, the age-dependent DNA methylation markers, as well as the mathematical models for age prediction and their evaluation. Furthermore, additional influences can have an impact. Therefore, the literature was evaluated in respect to these diverse topics.
Conclusion
The numerous research efforts in recent years have led to a rapid change in our understanding of the application of DNA methylation for chronological age prediction, which is now on the way to implementation and validation. Knowledge of the various aspects leads to a better understanding and allows a more informed interpretation of DNAm quantification results, as well as the obtained results by the age prediction tools.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Introduction
Within the last years, insights into the fascinating field of epigenetics increased in an expanse, which have also aroused attention in the field of forensic genetics. Until now, the use of epigenetically coded information of a trace found at a crime scene has not yet become a standard method in forensic casework laboratories. However, recent research demonstrates growing interest, and laboratories have started the development of assays for DNA methylation (DNAm) analysis, especially for tissue and body fluid identification (reviewed in An et al. 2012; Kader et al. 2020; Sijen and Harbison 2021). Additionally, research on the role of genomic imprinting for the determination of parent-of-origin alleles (Li et al. 1993; Zhao et al. 2005; Nakayashiki et al. 2009) or attempts for the authentication of DNA as biological material (Frumkin et al. 2010) was performed. Excellent reviews are available on the general usage of DNAm for forensic casework (Gršković et al. 2013; Vidaki et al. 2013; Gunn et al. 2014; Kader and Ghai 2015). The prediction of the chronological age of an individual became an intensely studied application using DNAm analysis. Many studies on age-dependent DNAm changes were often initiated by medical interest in the process of aging, including creation of epigenetic clocks (Teschendorff et al. 2010; Alisch et al. 2012; Horvath 2013; Hannum et al. 2013; Marioni et al. 2015). However, forensic scientists follow another goal, since prediction of chronological age is in focus compared to biological age and mortality risk. Furthermore, they deal with different types of challenging samples (e.g., low DNA quantity and quality), reproducibility and accuracy, as well as legal restrictions. The aim of the review is to provide an overview of concepts and considerations around markers, methods, models, and additional aspects for age prediction in the forensic setting due to available material to be analyzed (Fig. 1). The emphasis is also placed on the biological background, as age-dependent changes exist within the complex and dynamic framework of epigenetics.
Current methods and role of age prediction for forensic purposes
Forensic age prediction is a topic with long tradition and has been applied to narrow down the age of dead individuals to assist the identification of unidentified bodies (Ritz-Timme et al. 2000). Furthermore, age verification of living individuals is important for differentiation between legal age thresholds, playing, e.g., a role in case of immigration and in court (Schmeling et al. 2004, 2006). So far, age prediction is mainly based on the analysis of morphological and physiological characteristics, which are described and discussed elsewhere (Ritz-Timme et al. 2000; Schmeling et al. 2007). Alternatives based on molecular markers have been under investigation for a long time. These can be protein changes, i.e., accumulation of racemized aspartic acid and advanced glycation products (Brownlee 1995; Ritz-Timme and Collins 2002) or nucleic acid alterations, such as a 4977-bp mtDNA deletion (Lee et al. 1994), shortening of the telomere length (Tsuji et al. 2002; Takasaki et al. 2003; Karlsson et al. 2008), signal-joint T cell receptor excision circles (sjTRECs) (Zubakov et al. 2010; Ou et al. 2012; Cho et al. 2014), as well as changes in RNA expression (Peters et al. 2015). Age dependence for most of these markers is long known, however with limited potential for application due to low accuracy or dependence on specific tissues (Meissner and Ritz-Timme 2010). The potential of RNA as a biomarker for age prediction was demonstrated (Peters et al. 2015; Fleischer et al. 2018; Ren and Kuan 2020) but a possible application in a forensic setting needs to be further clarified.
Fundamentals of DNA methylation
For many years, the focus of forensic DNA analysis was restricted to the investigation of the ‘raw’ DNA sequence itself, primarily to determine the individual DNA profile using short tandem repeat (STR) analysis (Jeffreys et al. 1985b, a; Ellegren 2004; Jobling and Gill 2004). In addition to (almost) the same DNA sequence in all nucleated cells of a living organism, specific factors regulate our genome and thereby enable both cell-type-specific behavior and adaptations to internal as well as external influences (Feinberg 2001; Bjornsson et al. 2004; Boland et al. 2014). This concept was already proposed in 1942 by Waddington, shaping the term ‘epigenetics’ (reprint Waddington 2012). DNA methylation was proposed very early as a key factor in epigenetic regulation, first solely as an inhibitory regulator of gene expression (Riggs 1975; Holliday and Pugh 1975). However, today it is clear that regulation strongly depends on the location of methylation and thus can act both repressive and activating (Jones 2012). Other important representatives are packing structures (histones), regulatory DNA elements (e.g., enhancers), and noncoding RNAs (Goldberg et al. 2007).
DNAm (-CH3) in mammals occurs primarily at the fifth carbon atom of the base cytosine in the cytosine-guanine (CpG) sequence context. In addition, non-CpG dependent DNAm occurs but is restricted to neural and pluripotent cell types (Ziller et al. 2011; Arand et al. 2012). Hydroxymethylation (DNAhm) is a further modification regulating maintenance and differentiation of embryonic stem cells and present as an intermediate product during active removal of the methylation, respectively (Wallace et al. 2010; Hill et al. 2014; Zampieri et al. 2015).
Due to the double-stranded nature of DNA, DNAm occurs on both strands as 5′-CpG-3′ is also present on the opposite strand. Changes in the DNAm pattern occur mainly during development and cell differentiation, but are largely preserved in later cell divisions (Chen and Riggs 2011). In total, approximately 5% of cytosines are methylated (5mC) in the genome (referring to 80% of CpG positions being methylated). However, methylation is not evenly distributed throughout the genome with cell type and tissue-specific differences (Ehrlich et al. 1982; Ziller et al. 2013). In particular, long CpG-rich stretches, so-called CpG islands, contain a higher density of CpG sites (> 50%) and are mainly nonmethylated. Most of these regions (around 60%) are associated with promoters and nonmethylated sites are correlated to allow expression of the corresponding genes (Zampieri et al. 2015). In these regions, DNAm can block gene expression (Bird 1986; Cross and Bird 1995; Jones 2012), while DNAm in gene bodies can have the opposite effect (Razin and Riggs 1980; Jones 1999; Rauch et al. 2009; Lister et al. 2009; Laurent et al. 2010; Chen and Riggs 2011).
To what extent DNAm itself plays the decisive regulatory role or whether it mainly (permanently) stabilizes the epigenetic status given by histone modifications depends on the genomic location, the cell type, and time point (Jones 1999, 2012). Whereas the cell-type-specific DNAm patterns must be stable to preserve cell identity, a flexible change in DNAm can be generated by active methylation and demethylation. The latter can occur passively by loss of methylation or actively enzymatically, in particular via TET enzyme-based oxidation (Jones and Taylor 1980; Mayer et al. 2000; Oswald et al. 2000; Ma et al. 2009; Tahiliani et al. 2009; Wu and Zhang 2017). Through adaptations of DNAm to intrinsic and extrinsic changes, a stable and, at the same time, flexible chemical regulation is possible. Twin studies demonstrate that even if the genomic background is the same, epigenetic differences can be measured, caused by general epigenetic drift, as well as environmental differences (Fraga et al. 2005; Poulsen et al. 2007; Teschendorff et al. 2013; Issa 2014). Consistent with the dynamic side of DNAm, it has been shown that DNAm at some CpG positions changes in an age-dependent manner (Christensen et al. 2009; Teschendorff et al. 2010; Hannum et al. 2013).
Aging and age-dependent DNAm changes
Aging is a universal process that is at least partially controlled by genetic pathways and biochemical processes. During aging, physiological integrity decreases, leading to impaired functioning and thereby to an increased morbidity and mortality rate (Cevenini et al. 2008; López-Otín et al. 2013). Furthermore, there are individual and environmental differences (Melis et al. 2013). López-Otín et al. defined nine 'hallmarks of aging’ of which one is epigenetic alteration in addition to altered intercellular communication, stem cell exhaustion, cellular senescence, mitochondrial dysfunction, deregulated nutrient sensing, loss of proteostasis, telomere attrition, and genomic instability. Even if these hallmarks are labeled separately, they are interconnected, affecting each other, and therefore contributing together to the aging process and the resulting phenotype (López-Otín et al. 2013). Although we all undergo this process over time, large differences between individuals are observed in the phenotype of aging. That leads to differences in the biological ages between individuals who actually have the same chronological age (the period since birth). Multiple factors such as genetic background, environment, life style, and stochastic factors can be responsible for the observed differences (Candore et al. 2006).
The first indication of age-dependent changes in DNAm was found in the 1990s, revealing that altered and de novo DNAm can be observed in the promotor of IGF2 and the estrogen receptor in colon cells of aging individuals, as well as during cancerogenesis (Issa et al. 1994, 1996). In general, global hypomethylation is associated with age (Wilson and Jones 1983; Ca 1993; Bollati et al. 2009); however, local hypermethylation in CpG rich regions was identified (Maegawa et al. 2010; Rakyan et al. 2010; Bell et al. 2012). Genome-wide studies revealed high numbers of single age-dependent DNAm positions, of which some studies applied these to create mathematical models for age prediction (Fraga et al. 2005; Teschendorff et al. 2010; Koch and Wagner 2011; Horvath 2013; Hannum et al. 2013). These developments led to the additional terms 'epigenetic age' and 'epigenetic clocks', which refer to the measurement of a biological epigenetic marker (e.g., DNAm) and can contain information on the acceleration or deceleration of age in an individual by the difference between the measured epigenetic age and the chronological age (Horvath and Raj 2018). Some clocks were developed to explicitly predict biological age, including prediction of all-cause mortality (‘PhenoAge’, ‘GrimAge’, ‘DNAmFitAge’) (Levine et al. 2018; Lu et al. 2019; McGreevy et al. 2023). On the contrary, to create the chronological age prediction, markers have to be especially chosen based on their robustness to environmental factors, diseases, and phenotypes of an individual. Therefore, a good marker for chronological clock may not be a good candidate for biological age prediction, as good candidates were removed due to their high variation (correlated to biological variation) between individuals of the same age (Field et al. 2018). 'Robust' in this context does not mean that all possible altering conditions can be excluded, but that a prediction as robust as possible can be performed due to marker choice and use of a broad reference population trying to resemble the overall population (without selection for specific aspects such as smoking, nutrition behavior, and fitness). Therefore, division into ‘chronological epigenetic clocks’—also named ‘forensic age clock’—and ‘biological epigenetic clocks’ was proposed (Bell et al. 2019).
Ingredients and components to predict the chronological age
The idea behind forensic age prediction is to predict chronological age with the highest possible precision. First, an analysis method must be chosen that allows robust and reliable DNAm quantification, and can be applied in a forensic laboratory (‘DNA methylation analysis methods’). Second, age-dependent DNAm markers must be identified (‘DNAm markers’). Third, mathematical models must be created using a training set that covers a broad age range based on a large number of individuals (outbalancing a diverse spectrum of environment, diseases, etc.) and evaluated with test data not included in model development (‘Basics of age prediction models’). Furthermore, special aspects such as the tissue or body fluid type (‘Models developed for different tissues and body fluids’), amount of available DNA (‘Consideration of DNA amount’), and influences on DNAm (e.g., lifestyle, disease) (‘Other potential influences on the accuracy of age prediction’) should be considered with care.
DNA methylation analysis methods
DNAm does not change the DNA sequence itself. Therefore, it is not directly measurable via PCR and sequencing using standard approaches of forensic applications. During the PCR reaction, methylated cytosine is replaced with classical cytosine included in the PCR reaction mix, resulting in a loss of the DNAm pattern. The choice of the appropriate tool often depends not only on the task (e.g., qualitative versus quantitative analysis) but also on the amount of DNA, the optimization steps for the assay setup, the availability of analysis machines, and the costs, time, and expertise needed for processing of the samples. Therefore, only some methods will be highlighted here with an emphasis on methods used mainly in forensic epigenetics.
Three main categories of DNA pretreatment can be distinguished: (i) fixation of the DNAm pattern by bisulfite conversion; (ii) digestion of nonmethylated DNA by methylation sensitive DNA restriction enzymes (Bestor et al. 1984; Bickle and Krüger 1993; Huang et al. 1999); and (iii) selection of methylated DNA with the help of antibodies (MeDIP) (Weber et al. 2005). Although not largely present, it should be noted that these standard methods for DNAm detection cannot differentiate between DNAm and DNAhm, and the measured DNAm therefore contains the actual DNAm and the (rare) content of DNAhm.
Only bisulfite conversion will be explained in depth, as is the current gold standard for single base resolution of DNAm, and is the basis for the commonly used age prediction tools, for further reading about the other methods it is referred to (Harrison and Parle-McDermott 2011). Treatment with sodium bisulfite (sodium hydrogen sulfite) leads to sulfonation in pyrimidines (Hayatsu et al. 1970; Shapiro et al. 1973; Kai et al. 1974; Hayatsu 1976), which occurs much more slowly for methylated cytosines (Wang et al. 1980). Subsequent hydrolytic deamination and renewed desulfonation results in the formation of uracil at the positions of the originally nonmethylated cytosines. This chemical process leads depending on the kit used to some extent to DNA degradation and DNA loss (Holmes et al. 2014; Hong and Shin 2021). Bisulfite sequencing uses this approach in combination with the PCR and sequencing during which uracil is replaced by thymine (Frommer et al. 1992). Knowing by the reference sequence which position initially contained a cytosine, the DNAm status can be calculated by taking the amount of cytosine (initially methylated cytosines) divided by the amount of cytosine plus thymine (to uracil-converted nonmethylated cytosines) at that position. For analysis of the reverse strand, guanine and adenine must be considered for the calculation. For sequencing, common methods such as Sanger, massive parallel sequencing (MPS), pyrosequencing, and minisequencing (SNaPshot™) can be used. Especially pyrosequencing and MPS allow an exact quantification, as well as detection of multiple CpG sites for DNAm analysis and non-CpG sites for evaluation of the bisulfite conversion efficiency of the amplified fragments. Other possible methods also applied for DNAm-based age prediction are real-time PCR specific to methylation (Kondo et al. 2021), real-time PCR with high resolution melting (HRM) (Hamano et al. 2016, 2017), and digital droplet (ddPCR) (Shi et al. 2018; Han et al. 2020). In addition, the potential for DNAm analysis of nanopore sequencing was shown (Rand et al. 2017; Simpson et al. 2017). However, either they do not allow for the needed multiplex capacity (restricted by color channels in the case of real-time machines), enough resolution for an accurate single-based quantification, especially if multiple CpG sites are present in the amplicon (HRM), or they need specific equipment currently less used in forensic laboratories (ddPCR, Nanopore). Although not considered as a standard tool for forensic analysis, the Illumina Infinium microarray platform should not be neglected, as most of the markers today used were obtained using microarray data and the CpG ID numbering system (cg identifier) of Infinium microarrays. Three array types can be divided, 27 K, 450 K, and the EPIC version that covers more than 850,000 CpG sites (Bibikova et al. 2009, 2011; Pidsley et al. 2016). Since most of the data sets available online in recent years were based on the 450 K (and to a lesser extent the 27 K), these were often the basis for the selection of age-dependent DNAm markers (cf. ‘DNAm Markers’).
The choice of method also depends on the application. Currently, the two most important applications in forensic epigenetics are the differentiation between body fluids or tissues and the prediction of age. Although both are based on the measurement of DNAm patterns, the methodology used for age prediction must be more accurate, as changes of 1% of DNAm can be important (e. g., the mean increase in DNAm of the strong-changing marker ELOVL2 is less than 1% per year in middle-aged individuals (Naue et al. 2017)). In particular, to examine the variation obtained by the analysis methods, some studies have examined the differences that arise due to the technology used, and others have examined whether a reliable analysis is performed by different laboratories using the same technology (Freire-Aradas et al. 2020; Holländer et al. 2021; Naue et al. 2021a). Freire-Aradas et al. analyzed 84 blood samples with Epityper, pyrosequencing, MiSeq and minisequencing, gaining comparable results with the exception of minisequencing. Using a model based on all data from the four technologies, the highest discrepancies were identified for MIR29B2CHG (Freire-Aradas et al. 2020). Different studies have proposed approaches to account for such variation across technologies, such as including a variable that considers the used technology (Hong et al. 2019), applying a Z-score transformation (Feng et al. 2018; Freire-Aradas et al. 2020), or building specific models for each technology (Schwender et al. 2021).
Additionally, machine-type-specific differences can occur, as observed for minisequencing on the 3130 Genetic Analyzer and the newer 3500 model in collaborative exercises performed during the last years (Holländer et al. 2021; Naue et al. 2021a; Lee et al. 2022). So and Lee performed a deeper investigation, reanalyzing samples on the 3500 originally measured with the 3130 and concluded that the original age prediction model cannot be used and a new model was created for the 3500 (So and Lee 2021). Other studies also observed differences between the initial published results and their own implementations (Daunay et al. 2019; Pfeifer et al. 2020). Taken together, these results show the importance of solid verification during the implementation of published models in the own laboratory.
DNAm markers
The definition of ‘marker’ differs between publications and refers in the first place often to the genetic loci/sequence region and in the final model to the specific CpG position(s) analyzed. It has to be considered that a flexible biological marker (i.e., DNAm) is measured and that intra- and interindividual differences will occur even if especially markers for the purpose of chronological age prediction are selected. In the past, a large number of markers were identified in various studies based mainly on the determination of the Pearson’s product-moment correlation (r) and Spearman rank correlation (rho), respectively. Some of these markers are represented in multiple models, and were independently identified in studies generating own or using publicly available microarray data, or directly selected as potential candidates from the previous literature. Many markers were identified according to the selection criteria. Table 1 lists common markers incorporated in a final mathematical model for forensic application, at least used by two studies from different laboratories. Furthermore, markers applicable for age prediction in semen samples were also included if only mentioned once, (but partly validated in interlaboratory validation studies). However, markers were not included in the table when no model or only preliminary models without further evaluation were created due to the sample number, as, for example (Alsaleh et al. 2017; Naue et al. 2018b, 2021b; Lee et al. 2020). For the biological function of genes, it is referred to the NIH database (NCBI: https://www.ncbi.nlm.nih.gov/gene).
One of the first studies was conducted by Bocklandt et al. in 2011 selecting a small number of CpG positions applicable for forensic use. Initially, 88 age-dependent sites were considered in the study and narrowed to three loci (TOM1L1, EDARADD, and NPTX2), of which the final model included EDARADD and NPTX2 for the prediction of age in saliva (Bocklandt et al. 2011). The shortly after published Horvath clock included 353 CpG sites (Horvath 2013). When comparing the markers in Table 1 and the 353 CpG sites of the Horvath clock, only seven genes overlap with the loci currently used in forensic assays (KLF14, ITGA2B, LAG3, NOX4, PDE4C, RASSF5, and SCGN).
The probably best known and commonly used marker is ELOVL2, first mentioned in the publication by Garagnani et al. together with PENK and FHL2 (Garagnani et al. 2012). The fact that ELOVL2 does not overlap with Horvath’s clock is probably due to the lack of coverage of the corresponding CpG sites on the Illumina Infinium 27 K platform. Although the study also included 450 K data, only overlapping markers on both platforms were considered for the age prediction model (Horvath 2013). Multiple CpG sites in ELOVL2 correlate (nonlinear) with age over a wide range of tissues (cf. Table 1). However, tissue-specific characteristics, such as the amount of change per year and tissue-specific DNAm shift ('baseline DNAm') exist (Slieker et al. 2018; Naue et al. 2018b). Dependent on the study, only the CpG site that was most closely related or multiple highly correlating CpG sites were integrated into the published model of a study. Some markers were used in only a few studies (e.g., ARHGAP22, CNTNAP2, cg07082267, cg26947034, GRM2, NKIRAS2, F5, SYNE4), which may be due to multiple reasons. During marker selection, the question of the most suitable marker arises. Multiple thoughts have to be considered: (1) the choice of the correlation parameter, and threshold used for selection; (2) the number of markers included, depending on the analysis method and the model algorithm; (3) the amount of DNAm increase/decrease with age to facilitate differentiation also between low age ranges and to be higher than technical noise seems favorable; (4) the purpose of having a tissue-specific model or a cross-tissue model (e.g., ELOVL2 usable for age prediction based on multiple biological sources); (5) a marker as stable as possible also in case of disease or lifestyle/environment conditions. Furthermore, age-dependence of some markers is not always reproduced: ITGA2B showed only a weak correlation in blood (Bekaert et al. 2015b), but was identified before in other studies (Alisch et al. 2012; Weidner et al. 2014). Various issues such as the different age ranges covered, the tissue analyzed, the model used, and technical bias may be responsible for the observed differences. Furthermore, although the loci overlap between various studies, different CpG sites might have been used in the final model, as neighboring sites often correlate, leading to close Spearman correlation values, but might be slightly different between the studies. Furthermore, marker selection should not be handled too rigidly, as some markers might show a weaker correlation but could be useful for the reduction of outliers in a model, others might have a lower change with age, but show a very strong correlation with age (such as KLF14), and others might be very informative in specific tissues (such as semen). Therefore, a final assessment of the usefulness of a marker is difficult to do, and depends on the final aim (e.g., model specific for young age groups) and implementation (e.g., analysis method, mathematical algorithm).
Most research is conducted on autosomal gene regions but also the Y and X chromosome contain age-dependent CpG sites (Lund et al. 2020; Li et al. 2020b; Vidaki et al. 2021; Kananen and Marttila 2021; Jiang et al. 2023). Interestingly, different amounts of age-dependent sites were identified on the X-chromosome dependent on sex, with 1327 sites in men and only 325 sites in women, of which 122 CpG sites overlapped and five additional sites showed opposite age-dependent directions (Kananen and Marttila 2021). Especially, Y-chromosome-based age prediction would have advantages for forensic purposes, being male-specific and therefore usable in male–female DNA mixtures for age prediction of the male contributor. Lund et al. found between 40 and 169 Y chromosome CpG sites within four examined datasets, of which at least 82% of CpG sites showed hypermethylation, including seven CpG sites that overlapped the datasets (Lund et al. 2020). On the contrary, Kananen and Marttila found 46 age-dependent sites in at least two of the five analyzed datasets, but only two CpG sites overlapped in four of them. However, these two CpG sites did not overlap with the seven sites of Lund et al. That might be due to different age ranges, selection criteria, and technical differences (noise) between the data sets.
An additional interesting target for forensics would be DNAm in mtDNA (mtDNAm), as it would be especially useful in degraded samples that lack enough nuclear DNA. Controversial results were published, as different regions were analyzed, various methods applied for the analysis and the limited DNA conversion efficiency of circular DNA led to an overestimation of mtDNAm in some studies (Liu et al. 2016). A review by Cao et al. summarizes the observed difficulties and concludes that mtDNAm is on average between 1.5 and 5%, with some non-CpG sites reaching 10%, and has an asymmetric behavior (as the L-strand has a higher C-content) (Cao et al. 2021). Two recent studies detected age-dependent differences in postmortem brain tissue, with a low increase with age (< 10% DNAm), confirming the general low level of mtDNAm (Huang et al. 2022; Devall et al. 2023). The overall picture remains difficult, and more research is needed to get a deeper understanding of the potential of mtDNAm.
Basics of age prediction models
The selection of DNAm markers described above is the first step in model development. Age prediction models are trained using reference data (measured DNAm data and chronological age). The included DNAm markers are the initially selected features. Until now, a large number of age prediction models have been created. Most research groups created their age prediction model using one pre-selected algorithm, for example, multivariate linear regression (MLR) (Woźniak et al. 2021), multivariate quantile regression (Ambroa-Conde et al. 2022), random forest regression (Naue et al. 2017), and artificial neural networks (Vidaki et al. 2017). In other studies, multiple models were initially tested to select the best (Xu et al. 2015; Smeers et al. 2018; Aliferi et al. 2018; Freire-Aradas et al. 2022; Yang et al. 2023).
Independent of the model used, it is important that a training data set is used for model development and that independent data (not involved in model development) are used for the evaluation (Alzubi et al. 2018). In case of a low sample number, cross-validation (CV) methods (k-fold CV, leave-one-out (LOOCV)) can be a useful alternative. However, the use of an independent dataset, with an independent preparation of the samples, is advantageous for evaluating additionally intralaboratory batch effects between experiments. Interlaboratory exercises and validations would additionally allow the evaluation of batch effects between laboratories, as done in (Holländer et al. 2021; Naue et al. 2021a; Lee et al. 2022). During modeling, various things have to be considered. It is important to avoid an overfitted model, which can happen if the model parameters are chosen to perfect fit the training data set but are set too stringent for analysis of independent test data. This risk can be minimized by an initial feature selection using completely different data. Multiple studies realized this by using publicly available microarray data for initial marker selection, (among others Bocklandt et al. 2011; Weidner et al. 2014; Vidaki et al. 2017; Naue et al. 2017; Freire-Aradas et al. 2022).
Model evaluation is performed primarily using the mean absolute error (MAE), or the root mean square error (RMSE) (Handelman et al. 2019). Furthermore, the median absolute evaluation can be used, which is also sometimes abbreviated MAE, and should not be mixed with the mean absolute error when comparing models (therefore, abbreviated MedAE in this review). Additionally, the percentage of correct predictions within in an acceptable error range (mainly ± 5 years) is often stated as in (Zbieć-Piekarska et al. 2015a; Pan et al. 2020; Freire-Aradas et al. 2022). All these values are useful for evaluating the overall model; however, they do not provide information about the maximum observed deviation between predicted and actual age, nor about the confidence in a single prediction.
Comparisons between models based on their MAE or RMSE should be made with caution. An increased range between DNAm values from individuals of the same age was measured especially in the elderly (Fraga et al. 2005; Martino et al. 2013) and was also measurable by the increasing MAEs in age group-specific analyses (Bekaert et al. 2015b; Naue et al. 2017). Therefore, a model that covers a wide age range, and is tested in many older individuals can result in a worse overall MAE compared to a model tested with a larger dataset of young individuals. Furthermore, different models are based on various needs and therefore compromise, e.g., ease of implementation in a forensic laboratory, number of analyzable markers, accurate predictions for a specific age range, tissue specificity, and need of a universal approach, respectively.
Models developed for different tissues and body fluids
Age-dependent DNAm changes are tissue-specific and must be considered (Day et al. 2013; Slieker et al. 2018). Although the Horvath epigenetic clock was created as a universal clock, it clearly shows differences in prediction accuracy between tissues with an overall MedAE of 3.6 years in the overall test data, with 3.7 years for whole blood, but 18 years for skeletal muscle (Horvath 2013). Therefore, many studies have been conducted to identify age-dependent markers specific for a tissue or to adapt the model to the reference data for each tissue. Table 1 does not claim to provide all available studies, as far more were published, and in addition yet unpublished and modified models exist, respectively. Many models developed for forensic purposes show an MAE (often referred as accuracy) of 3–5 years (cf. sections below). An overview of forensically motivated studies is provided below for analysis of different types of tissue commonly encountered in criminal investigations.
Blood
Initially, most of the studies were developed and optimized for DNA analysis from blood (among others Weidner et al. 2014; Zbieć-Piekarska et al. 2015a, b; Huang et al. 2015; Bekaert et al. 2015a, b; Park et al. 2016; Freire-Aradas et al. 2016, 2022; Thong et al. 2017; Vidaki et al. 2017; Cho et al. 2017; Naue et al. 2017; Aliferi et al. 2018, 2022; Jung et al. 2019; Daunay et al. 2019; Alsaleh and Haddrill 2019; Han et al. 2020, 2022; Correia Dias et al. 2020b). Zbieć-Piekarska et al. developed one of the first models with an MAE of 3.9 years (Zbieć-Piekarska et al. 2015a). The set of markers analyzed (ELOVL2, FHL2, KLF14, TRIM59, and MIR29B2CHG (initially named C1orf132)) is the most investigated set. It was further evaluated applying other mathematical algorithms, using other methods and populations (Cho et al. 2017), and in studies investigating the effect of diseases (Spólnicka et al. 2016, 2018b, c). Furthermore, these markers were also independently identified and/ or implemented by other studies included in this review (Table 1).
Blood samples from deceased individuals were also examined and no generally biased DNAm results have been found so far (Hamano et al. 2016; Naue et al. 2018b; Correia Dias et al. 2020a; Pfeifer et al. 2020). However, these studies did not contain a detailed systematic investigation of the effect of different stages of putrefaction. Another possible point to consider is the cell type composition, whose role for DNAm analysis was examined in larger microarray studies and normalization procedures were developed (Houseman et al. 2012; Teschendorff et al. 2017). However, these methods require the use of microarray data for cell type deconvolution and are therefore not suitable for forensic purposes. Although the observed changes in DNAm may be correlated with changes in cell type with aging, these do not necessarily interfer with age prediction but can refer to markers highly specific to blood. Jaffe and Irizarry analyzed the association of blood cell composition and DNAm and found loci with statistically significant different DNAm depending on cell type count (including the well-known FHL2) (Jaffe and Irizarry 2014).
Studies using age-dependent sites on the Y chromosome have also been conducted so far only in blood. Vidaki et al. developed a support vector machine (SVM) radial model, resulting in an MAE of 7.54 years (75 CpG sites) and 8.46 years (reduced selection of 19 CpGs) for the validation set. Interestingly, in contrast to autosomal age prediction, Y-based prediction did not increase in the elderly (Vidaki et al. 2021). Very recently, Jiang et al. developed an age prediction model based on minisequencing and random forest regression (including 13 CpGs) with an MAE of 5.73 years in the test set, including individuals between 21 and 100 years (Jiang et al. 2023).
Saliva and buccal cells
Saliva is a common trace material and the first material on which an age prediction model with an MAE of 5.2 years (LOOCV) was established for forensic purposes (Bocklandt et al. 2011). Saliva was investigated in further studies using the same markers as for blood-based models and/ or markers specific for saliva (Hong et al. 2017; Hamano et al. 2017; Jung et al. 2019; Ambroa-Conde et al. 2022). Further studies used buccal swabs as tissue source (Bekaert et al. 2015a; Pfeifer et al. 2020; Han et al. 2020; Woźniak et al. 2021; Schwender et al. 2021). Although not a trace material, buccal swabs are often used (and are easily applicable) in research studies and could be useful for age verification in living individuals.
However, buccal cell swabs and saliva cannot be considered as interchangeable material for epigenetic analysis. Saliva is a heterogeneous body fluid with a mixture of leukocytes and epithelial cells of the oral cavity. As a result, the composition of cell types may be more related to blood or to buccal swabs. However, also a buccal swab is heterogeneous, as leukocytes are also obtained during sample collection (Theda et al. 2018). To account for these heterogeneous materials, cell-type-specific markers can be included to determine the epithelial/leukocyte ratio. In one study, CpG sites in CD6 (cg07380416) and SERPINB5 (cg20837735) DNAm were analyzed obtaining a ‘Buccal Cell Signature’ that together with the 3-CpG age prediction set improved age prediction by decreasing the MAE from 7.03 years (3-CpG model) to 5.09 years (5-CpG model) in the independent validation set (Eipel et al. 2016). In another study, the authors were able to improve their model by analyzing a cell type-specific CpG site in PTPN7 (cg18384097), resulting in an MAE of 3.15 years in contrast to the MAE of 4.1 years without PTPN7 inclusion. In particular, they were able to reduce partially the deviation from chronological age in the elderly group (Hong et al. 2017). The genes PTPN7 (coding the nonreceptor protein tyrosine phosphatase type 7) and CD6 (a T cell differentiation gene) show hypermethylation in epithelial cells and hypomethylation in blood cells, while SERPINB5 (serpin peptidase inhibitor clade B member 5) shows the opposite DNAm pattern with hypomethylation in epithelial cells and hypermethylation in blood (Eipel et al. 2016; Hong et al. 2017). Ambroa-Conde et al. used another approach considering the difference between saliva and buccal swab material, developing a combined minisequencing assay for buccal cells and saliva, resulting in a MedAE of 3.66 years. Furthermore, they tested the inclusion of CpG sites in HUNK and RUNX1 to predict the tissue source as an additional covariable for the age prediction tool, which did not improve age prediction but led to correct classification of the tissue source in 83.69% of the cases (Ambroa-Conde et al. 2022).
Two studies were able to analyze DNAm markers of buccal swab material from deceased individuals (Naue et al. 2018b; Koop et al. 2021). Koop et al. investigated whether the decomposition stage has an influence on PDE4C DNAm. No dependence on the decomposition stage was found as long as enough DNA was recovered. Furthermore, no association was detected between DNA degradation until the decomposition stage six, at which tissues started to dry out. A higher amount of buccal cells and DNA was even obtained in cases of mid-level decomposition, which the authors speculate could be due to decreased mucosal stability and, therefore, easier cell collection (Koop et al. 2021).
Semen
Unlike somatic tissues that show global hypomethylation and regional hypermethylation, in semen the opposite trend can be observed, and using the mean DNAm values of 51 regions analyzed by microarray, Jenkins et al. were able to construct an MLR model with an MAE of 2.37 for the ten independent test samples (Jenkins et al. 2018). However, only a few studies have been conducted with emphasis on forensic applications, including an exploratory study and a validation by Lee et al. in 2015 and 2018. In these studies, one CpG in TTC7B, cg12837463, and NOX4, respectively, were analyzed by minisequencing to obtain an MAE of 4.8 years (Lee et al. 2015, 2018). Li et al. also included TTC7B and NOX4 in their age prediction tool using pyrosequencing and linear regression to obtain an MAE of 4.16 in fresh samples and 4.39 years in aged semen samples (Li et al. 2020a). Furthermore, the VISAGE consortium selected age-dependent sites in semen and conducted interlaboratory studies to verify the robustness of DNAm obtained from 13 markers (Heidegger et al. 2022). The final model is based on the MPS analysis of 6 CpG sites in five genomic regions (SH2B2, EXOC3, GALR2, IFITM2, FOLH1B) and led to an MAE of 5.1 years (RMSE 6.3 years) using an MLR model (Pisarek et al. 2021).
Hair
Hair is often found at crime scenes, therefore two studies have investigated age-dependent changes in the hair follicles of dead or living individuals using MPS (Naue et al. 2021b) or minisequencing (Hao et al. 2021), respectively. In the case of the latter study, an MLR prediction model comprising 10 CpG sites was developed and an MAE of 4.15 years (RMSE 4.92 years) in the test set was obtained. No correlation with sex or hair color was found (Hao et al. 2021), but the plucked hair follicle can be very heterogeneous in the amount of DNA obtained, which is an important factor for successful DNAm analysis (Naue et al. 2021b). Further studies are needed and the application has its limits, as hairs found at crime scenes are mainly telegenic hairs or only hair shafts.
Bone
So far, some studies have investigated age-dependent DNAm in bones (Shi et al. 2018; Naue et al. 2018b; Gopalan et al. 2019; Lee et al. 2020; Woźniak et al. 2021; Becker et al. 2021; Correia Dias et al. 2021). In a larger study, Woźniak et al. included, in addition to other types of tissue, 161 bone samples from the occipital bone or femoral shaft, of which 112 were used for model training and 49 for testing. Two selected sites in ELOVL2 and PDE4C each, as well as one CpG site in KLF14 and ASPA each, were analyzed by MPS and implemented in a bone-specific MLR model. The obtained MAE of 3.4 years was comparable to the results from blood with the blood-specific model developed in the same study (MAE of 3.2 years). They also tested whether their bone model could be used to predict the age of other types of tissue. Age prediction of blood samples resulted in a quite good estimate with an MAE of 4.9 years, while cartilage and muscle predictions were much less accurate with 25.8 and 13.7 years, respectively (Woźniak et al. 2021).
Teeth
Another material with potential to predict the age of especially deceased individuals are teeth samples. Several studies have been conducted so far investigating markers also identified in other tissue types (among others ELOVL2, FHL2, EDARADD, and PDE4C). These confirmed the presence of age-dependent changes also in teeth and the potential to build models for age prediction (Bekaert et al. 2015b; Márquez-Ruiz et al. 2020; Kondo et al. 2021; Zapico et al. 2021; Correia Dias et al. 2021). Interestingly, Giuliani et al. found that the part of the tooth as a source of DNA plays an important role. Not only the amount of DNA obtained, but also the accuracies of the age predictions were different. Dental pulp material resulted in slightly better results (MedAE 2.25 years) than cementum (MedAE 2.45 years), while age prediction from dentin samples was the least accurate (7.07 years). The best result with an MedAE of 1.2 years was obtained by a combination of pulp and cementum. The authors speculate that the differences could be due to different types of dentin during life, with tertiary dentin (secreted in response to external damage) being different between individuals leading to a greater variability in age prediction, especially for the elderly (Giuliani et al. 2016). However, further evaluation is needed with additional independent samples.
Consideration of DNA amount
There is a general need to obtain a reliable result for the analysis of trace material, including small and degraded amounts of DNA. However, in case of DNAm analysis, an additional layer of variation for the successful outcome has to be considered since the DNAm at one CpG site of one DNA molecule is a bivariate characteristic (methylated versus unmethylated). The DNAm level between 0 and 100% represents the DNAm of a cell population and, therefore, over a number of different cells. As traces often contain only a few cells of this population, different traces from the same original source can contain different DNAm patterns (Naue et al. 2018a). It should be noted that the DNAm still presents the biological value of the few analyzed cells, but not necessarily the overall tissue DNAm of an individual. Furthermore, the final measured deviation also contains technical variation. The influence of the DNAm variation on the age prediction model has to be evaluated, and might vary depending on markers, models, and the investigated age range. Therefore, multiple studies investigated the sensitivity threshold for their own assays and markers, with varying results between 1 and 20 ng (Zbieć-Piekarska et al. 2015b; Hong et al. 2017; Heidegger et al. 2020, 2022; Woźniak et al. 2021; Aliferi et al. 2022). However, these results cannot be directly compared, as the terms ‘sensitivity’, ‘robust’, and ‘reliable’ do not have a consistent definition for DNAm analysis and age prediction. Additionally, there are differences in the setup of the assay (elute volume from bisulfite conversion used for PCR, analysis methods, tissue type, and fragment sizes). Woźniak et al. were able to robustly quantify the DNAm of most markers down to 20 ng DNA input for bisulfite conversion, referring to approximately 8.8–11.8 ng converted DNA in the PCR (Woźniak et al. 2021). The same research consortium obtained stable normalized read depth and accurate DNAm results for all markers at 50 ng input for conversion and a possible 11–14.8 ng input in PCR for the analysis of semen samples (Heidegger et al. 2022). Within the same range, the results of other studies that developed an age prediction assay in buccal cells and saliva, leading to a minimum amount of 10 ng (Ambroa-Conde et al. 2022). An increased absolute error in the age prediction between duplicates or triplicates was also obtained in other studies using down to 2.5 ng and 1 ng, respectively in a pyrosequencing and MPS assay (Zbieć-Piekarska et al. 2015b; Heidegger et al. 2020). However, Aliferi et al. were able to still get reproducible results down to 5 ng of DNA input for bisulfite conversion, resulting in 1 ng of converted DNA for PCR, demonstrating possible high sensitivity for age prediction (Aliferi et al. 2022). Using minisequencing, a technical limit can be the appearance of allelic dropout, resulting in a threshold of 4 ng and 5 ng of converted DNA for two assays investigating saliva and semen samples (Hong et al. 2017; Lee et al. 2018). Jiang et al. also evaluated the sensitivity for their Y-based assay. A complete electropherogram was obtained down to 0.5 ng (Jiang et al. 2023).
Other potential influences on the accuracy of age prediction
Epigenetic modifications form a regulatory layer and can be influenced by genetics and environmental effects. The attribution of each factor is especially observable in twin studies (Fraga et al. 2005; van Dongen et al. 2016; Hannon et al. 2018; Reynolds et al. 2020). Therefore, its influence on age-dependent markers must be investigated. As many factors may contribute to the variation, only some studies can be highlighted. Although divided into sections, influences can be interconnected and may cover other smaller effects.
Biological sex
Generally, sex-dependent differences in DNAm have been reported (Fuke et al. 2004; El-Maarri et al. 2007; Boks et al. 2009; Hannum et al. 2013; Marttila et al. 2013; Zaghlool et al. 2015). Therefore, a possible biological sex dependence on age prediction was considered since the development of the first forensic age prediction models. No significant sex-dependent differences were found in most studies developing prediction models (Koch and Wagner 2011; Bekaert et al. 2015b; Eipel et al. 2016; Freire-Aradas et al. 2016; Vidaki et al. 2017; Aliferi et al. 2022). Non-significant tendencies were observed in other studies (Zbieć-Piekarska et al. 2015a; Naue et al. 2017). Since there may be small differences depending on the markers and models chosen, future models should also be checked for differences due to sex, but, so far, it can be concluded that if sex is having an influence, then the impact on the accuracy of age prediction is rather low.
Reference population
Furthermore, the reference population of the model has to be considered. General population-dependent DNAm differences were found by investigating genome-wide DNAm profiles of different populations, which are caused by genetic and environmental differences (Fraser et al. 2012; Heyn et al. 2013; Gopalan et al. 2017; Carja et al. 2017). Most age-dependent markers were initially identified by a combination of various publicly available data sets covering various worldwide populations (Horvath 2013; Hannum et al. 2013; Vidaki et al. 2017; Naue et al. 2017; Aliferi et al. 2022). However, the available data sets are not evenly distributed, and not all geographic regions are covered by these studies. Furthermore, many final prediction models were created on samples collected at the geographic location of the research laboratories that performed the study, e.g., residents of the Netherlands (Naue et al. 2017), the UK (Aliferi et al. 2022), South Korea (Hong et al. 2017), Poland (Zbieć-Piekarska et al. 2015a), and Germany (Schwender et al. 2021). Although not specified in detail, it can be assumed that the biogeographic ancestry of some of the individuals may be different, as well as the period of residence (and therefore environmental exposure duration) at the sampling location.
Various studies included investigations of DNAm differences between populations or validated published models for applicability in another population than the one included for model development (Eipel et al. 2016; Vidaki et al. 2017; Cho et al. 2017; Fleckhaus et al. 2017; Daunay et al. 2019; Aliferi et al. 2022). The results obtained are heterogeneous, since no differences were obtained in some studies (Eipel et al. 2016; Vidaki et al. 2017; Aliferi et al. 2022), while differences were observed in other studies (Cho et al. 2017; Gopalan et al. 2017; Fleckhaus et al. 2017; Becker et al. 2022). However, these differences were versatile. Cho et al. obtained a consistent general performance of the model analyzing South Korean individuals with a model based on individuals from Poland. However, the degree of age correlation showed differences, resulting in a retraining of the model for further improvement (Cho et al. 2017). Another study saw different amounts of interindividual variation for ELOVL2 depending on ancestry (Fleckhaus et al. 2017), while Daunay et al. validated six existing models based on other populations for their prediction accuracy in a French population, obtaining a general lower model accuracy for some of the models (Daunay et al. 2019). So far, a generalization is not easily possible, as observed differences could be specific for the investigated population, being partly a technical and/or sample batch effect, be caused by a different setup (marker, age range, analysis method), and a combination of all factors, respectively. Another issue to consider is that the reference age is provided by the participants. Gopalan et al. accounted for the role and difficulties of age verification for specific populations by including birth and wedding certificates, school records, local and historical events, and other forms for cross-verification (Gopalan et al. 2017).
Environmental exposures and lifestyle
The influences of various environmental exposures, such as air pollution, lead, mercury, or bisphenol A, on genome-wide DNAm changes has been observed in various studies and were summarized and discussed by Martin and Fry in 2018 (Martin and Fry 2018). Lifestyle factors such as smoking and alcohol use disorder are also often associated with changes in DNAm patterns as reviewed in (Zahs et al. 2012; Lee and Pausova 2013; Zhang and Gelernter 2017; Kaur et al. 2019; Zong et al. 2019). Regarding the forensic age prediction tools developed, Aliferi et al. found no impact on their prediction model due to environmental and lifestyle differences between the sampled individuals from the UK and Spain (Aliferi et al. 2022). Eipel et al., as well as Schwender et al., specifically analyzed the effect of smoking, and found no smoking-associated differences in their buccal swab samples at 5 CpG sites in ASPA, ITGB2B, PDE4C, CD6 and SERPINB5, and 88 CpG sites in the loci of PDE4C, ELOVL2, ITGA2B, ASPA, EDARADD, SST, KLF14, SLC12A5, respectively (Eipel et al. 2016; Schwender et al. 2021). Piniewska-Róg et al. investigated DNAm changes at 44 CpG sites in ASPA, EDARADD, ELOVL2, FHL2, KLF14, MIR29B2CHG, PDE4C, and TRIM59 in deceased extensive alcohol abusers. However, only an effect was observed in MIR29B2CHG, without a relevant impact on the age prediction using the blood-based VISAGE enhanced age model (including CpGs in ELOVL2, FHL2, KLF14, MIR29B2CHG, and PDE4C) (Piniewska-Róg et al. 2021).
The influence of extreme sport was also investigated as possible impact. Age predictions of elite athletes resulted in increased predicted ages, especially caused mainly by KLF14 and TRIM59. The effect was more pronounced (especially due to changes in TRIM59) for men and women who perform power sports (Spólnicka et al. 2018a).
Disease and medical treatment
Furthermore, DNAm changes associated with a disease can lead to changes in the precision of age prediction. Some diseases might be the result of an environmental exposure or lifestyle; thus, a changed DNAm may be the result of the lifestyle as well as the disease (cf. excessive alcohol consumption covered before). The following examples show the complex nature and possible impact of this topic.
For example, in the case of chronic lymphocytic leukemia, age in patients was not correctly predicted anymore using the markers ELOVL2, MIR29B2C, TRIM59, KLF14, and FHL2 (Spólnicka et al. 2018c). Spólnicka et al. also investigated the effect of allogeneic hematopoietic stem cell transplantation on the prediction of recipient age. They found that the measured age of the recipient after transplantation was more correlated with the chronological age as well as the with the age prediction model calculated age of of the donor than with the chronological age of the recipient, confirming the observations of Weidner et al. (Weidner et al. 2015). However, Spólnicka et al. observed with their age prediction model a mean recipient age prediction that was 3.7 years lower than that of the donor, while the prediction in the Weidner et al. study was 7 years higher than the chronological age of the donor. The authors found the reason for the lower predicted age in hypermethylation of MIR29BCHG2 (the higher the DNAm of MIR29BCHG2, the lower the predicted age), a marker that is not present in the model of Weidner et al. (Weidner et al. 2015; Spólnicka et al. 2016). Both studies had a maximum period of one year. Therefore, it might be interesting to see whether the observed effects are constant or change over a longer period. Spólnicka et al. questioned whether there is a dissociation of the age dependency of MIR29BCHG2 and specific circumstances, and a rethinking of the usefulness of this marker for forensic age prediction would be needed (Spólnicka et al. 2016). However, they also found in another study that this marker is stable in other diseases, while TRIM59 and KLF14 showed hypermethylation in early-onset Alzheimer’s disease, and hypermethylation of TRIM59 and hypomethylation of FHL2 in Graves’ disease (Spólnicka et al. 2018b). Although single-marker-based predictions led to large discrepancies between chronological and prediction age (up to 10 years), using the original 5-marker age prediction model, an increase of only 1.7 years was observed for early-onset Alzheimer's disease in the entire age range (6 years in the younger group). No bias was found in the case of Graves’ disease, which can be explained by the opposite changes in DNAm of TRIM59 and FHL2 (Spólnicka et al. 2018b). Having a look at various diseases, Aliferi et al. found no bias due to schizophrenia, rheumatoid arthritis, frontal temporal dementia, and progressive supranuclear palsy for their 11-marker model based on blood. However, at the gene level, they conclude that there are potential associations with obesity, smoking, metabolic, and cardiovascular diseases. This does not directly lead to an effect on the accuracy of age prediction, but associations between age markers and key changes during aging must be considered (Aliferi et al. 2022).
These examples reinforce the advantage of using multiple markers for age prediction. The overall observed differences of the impact on age prediction are not surprising as the markers analyzed as well as the tissue source investigated will be affected differently in case of an underlying disease.
Consequences
Information about the biological sex of a person can be determined during standard STR-profiling and could therefore be easily considered if needed. On the contrary, biogeographic ancestry analysis is currently limited to classification of continental regions of East Asia, South Asia, Europe, sub-Saharan Africa, Oceania, America (indigenous population) and is restricted due to legal restrictions in various countries (Schneider et al. 2019). Although the results have been controversial, knowledge of the biogeographical ancestry of the trace causer would still be beneficial for forensic application. The investigator could be more cautious with the interpretation of the result obtained. Nevertheless, the determination about the biogeographic ancestry does not allow conclusion about the residence location of the individual and therefore environmental exposures. Furthermore, the background on lifestyle and disease will not be known in the case of trace material (assumptions might be possible if a trace is found in connection with a specific lifestyle, for example, on a cigarette butt) and could only be considered if an age verification of a living individual is required. Common environmental exposures, as well as lifestyles such as smoking and moderate alcohol consumption will be covered in most prediction tools; as a part of the individuals included for training will also have this lifestyle. Knowledge of disease in the reference data is more complex, even if a ‘healthy control group’ was included. General predispositions, unknown diseases, and conditions related to the aging process will still be included in the assays and be present in the individual who left the analyzed trace.
Combined use of methods for age prediction
Although this review focuses on age prediction using DNAm changes, the combination of biomarkers could be a useful approach. The use of a second estimator could verify or question the predicted age by DNAm analysis. So far, only a few studies have investigated this potential. Márquez-Ruiz et al. found no relevant improvement in age prediction accuracy when telomere length was combined with DNAm in a small study that examined teeth (Márquez-Ruiz et al. 2020). Zubakov et al. compared the potential of mRNA, sjTREC, telomere length and DNAm and confirmed that the highest precision was due to DNAm analysis, but that mRNA provided additional independent information useful for a combined analysis (Zubakov et al. 2016). Another example is the combination of DNAm and sjTREC as done by Cho et al. obtaining an improved prediction for the elderly (Cho et al. 2017). Other studies investigated the combination of different kinds of age-dependent changes such as the combination of skeletal, dental age, and DNAm in children (Shi et al. 2018) or the idea of combining age-dependent protein changes and DNAm (Becker et al. 2021). Even if only small improvements might be observed by adding an additional layer to the prediction, the identification of outliers and perhaps a better age prediction of these would also be an important improvement for forensic purposes.
Mammalian age prediction
As aging is not restricted to the human species, age prediction can also be performed for other types of animal. Lu et al. constructed three universal pan-mammalian clocks using cytosines in highly conserved DNA stretches of 185 mammalian species (19 taxonomic orders) including 59 tissue types and an age range from prenatal to 129 years. As seen in humans, age-dependent cytosines are enriched at polycomb sites. The basic clock included all available animals without adaptions to different species conditions, whereas a normalized universal relative age clock considered the maximum lifespan, and a third clock normalized to sexual maturity and gestation time. Each clock was built from fewer than 1,000 CpG sites, and the chronological age versus the predicted age showed a median error of less than a year. However, species-specific differences occurred, and a lower correlation was achieved for example for bowhead whales (Lu et al. 2021). First age prediction clocks for specific animals with potential forensic relevance were also developed, including horses, dogs, cats, elephants, and apes (Ito et al. 2018; Prado et al. 2021; Horvath et al. 2021, 2022a, b; Raj et al. 2021).
Conclusions and outlook
This review aimed to provide an overview of DNAm analysis for age prediction, including various aspects to consider. Often it was only possible to provide examples. The application of DNAm for age prediction is reasonable and multiple tools were developed. In the future, the focus should be on identifying the sources that lead to outliers to optimize the models by or to identify a possible prediction outlier. Multimarker models seem favorable for that, outbalancing single DNAm changes, and facilitating outlier detection. However, the number of markers should be within the range of the possibilities of multiplex PCRs, avoiding also the need for too much DNA.
Although most of the models referred to were developed for one type of tissue, some were developed to allow the analysis of multiple tissues. Different approaches are possible: (i) analysis of the same markers, but in part use and considering different CpG sites within different tissue-specific models as in the enhanced age prediction panel of the VISAGE consortium for age prediction using blood, buccal cells or bone material (Woźniak et al. 2021), (ii) measurement and use of the same CpG sites but with tissue-specific models as done by Jung et al. in the minisequencing assay for age prediction using blood, buccal cells, and saliva (Jung et al. 2019), and (iii) application of a universal model including various tissues and cell types (Horvath 2013). Although a universal model appears favorable, the accuracy of the model varied depending on the tissue type, and therefore, tissue-specific models based on one flexible experimental assay as used in (i) and (ii) currently are more promising for application in forensic investigations.
To allow for better comparability between studies in the future, the use of CpG sites present on microarrays might be favorable because they are investigated in a large number of studies. Although the best CpG site may have varied between studies, this phenomenon needs to be further investigated, as the sample size and batch effects can result in small differences in the correlation values. As various analysis methods will be used in the future, normalization strategies or methods allowing reliable machine-independent absolute quantification are needed. Furthermore, the application for trace material has not yet been fully evaluated. More studies on DNA quantification with respect to DNA quantity and quality, as well as changes caused by trace exposure to factors such as ultraviolet light and the time since deposition, must be performed. In the case of deceased individuals, more information on postmortem stability is needed for more tissues. Furthermore, not all types of trace material were investigated to the same depth, as for some only first attempts were made (e.g., menstrual blood, vaginal secretion in (Alsaleh et al. 2017)). In particular, more interlaboratory exercises will help to further optimize and implement age prediction tools in forensic laboratories (Holländer et al. 2021; Naue et al. 2021a; Heidegger et al. 2022; Lee et al. 2022).
References
Alghanim H, Antunes J, Silva DSBS, Alho CS, Balamurugan K, McCord B (2017) Detection and evaluation of DNA methylation markers found at SCGN and KLF14 loci to estimate human age. Forensic Sci Int Genet 31:81–88. https://doi.org/10.1016/j.fsigen.2017.07.011
Aliferi A, Ballard D, Gallidabino MD, Thurtle H, Barron L, Syndercombe Court D (2018) DNA methylation-based age prediction using massively parallel sequencing data and multiple machine learning models. Forensic Sci Int Genet 37:215–226. https://doi.org/10.1016/j.fsigen.2018.09.003
Aliferi A, Sundaram S, Ballard D, Freire-Aradas A, Phillips C, Lareu MV, Court DS (2022) Combining current knowledge on DNA methylation-based age estimation towards the development of a superior forensic DNA intelligence tool. Forensic Sci Int Genet 57:102637. https://doi.org/10.1016/j.fsigen.2021.102637
Alisch RS, Barwick BG, Chopra P, Myrick LK, Satten GA, Conneely KN, Warren ST (2012) Age-associated DNA methylation in pediatric populations. Genome Res 22(4):623–632. https://doi.org/10.1101/gr.125187.111
Alsaleh H, Haddrill PR (2019) Identifying blood-specific age-related DNA methylation markers on the Illumina MethylationEPIC® BeadChip. Forensic Sci Int 303:109944. https://doi.org/10.1016/j.forsciint.2019.109944
Alsaleh H, McCallum NA, Halligan DL, Haddrill PR (2017) A multi-tissue age prediction model based on DNA methylation analysis. Forensic Sci Int Genet Suppl Ser 6:e62–e64. https://doi.org/10.1016/j.fsigss.2017.09.056
Alzubi J, Nayyar A, Kumar A (2018) Machine learning from theory to algorithms: an overview. J Phys: Conf Ser 1142:012012. https://doi.org/10.1088/1742-6596/1142/1/012012
Ambroa-Conde A, Girón-Santamaría L, Mosquera-Miguel A, Phillips C, Casares de Cal MA, Gómez-Tato A, Álvarez-Dios J, de la Puente M, Ruiz-Ramírez J, Lareu MV, Freire-Aradas A (2022) Epigenetic age estimation in saliva and in buccal cells. Forensic Sci Int Genet 61:102770. https://doi.org/10.1016/j.fsigen.2022.102770
An JH, Shin K-J, Yang WI, Lee HY (2012) Body fluid identification in forensics. BMB Rep 45(10):545–553. https://doi.org/10.5483/bmbrep.2012.45.10.206
Arand J, Spieler D, Karius T, Branco MR, Meilinger D, Meissner A, Jenuwein T, Xu G, Leonhardt H, Wolf V, Walter J (2012) In vivo control of CpG and non-CpG DNA methylation by DNA methyltransferases. PLoS Genet 8(6):1002750. https://doi.org/10.1371/journal.pgen.1002750
Becker J, Böhme P, Reckert A, Eickhoff SB, Koop BE, Blum J, Gündüz T, Takayama M, Wagner W, Ritz-Timme S (2022) Evidence for differences in DNA methylation between Germans and Japanese. Int J Legal Med 136(2):405–413. https://doi.org/10.1007/s00414-021-02736-3
Becker J, Naue J, Reckert A, Böhme P, Ritz-Timme S (2021) Nutzung von Altersinformationen aus posttranslationalen Proteinmodifikationen und DNA-Methylierung zur postmortalen Lebensaltersschätzung. Rechtsmedizin 31:234–242. https://doi.org/10.1007/s00194-021-00489-2
Bekaert B, Kamalandua A, Zapico SC, Van de Voorde W, Decorte R (2015a) A selective set of DNA-methylation markers for age determination of blood, teeth and buccal samples. Forensic Sci Int Genet Suppl Ser 5:e144–e145. https://doi.org/10.1016/j.fsigss.2015.09.058
Bekaert B, Kamalandua A, Zapico SC, Van de Voorde W, Decorte R (2015b) Improved age determination of blood and teeth samples using a selected set of DNA methylation markers. Epigenetics 10(10):922–930. https://doi.org/10.1080/15592294.2015.1080413
Bell JT, Tsai PC, Yang TP, Pidsley R, Nisbet J, Glass D, Mangino M, Zhai G, Zhang F, Valdes A, Shin SY, Dempster EL, Murray RM, Grundberg E, Hedman AK, Nica A, Small KS, MuTHER Consortium, Dermitzakis ET, McCarthy MI, Mill J, Spector TD, Deloukas P (2012) Epigenome-wide scans identify differentially methylated regions for age and age-related phenotypes in a healthy ageing population. PLoS Genet 8(4):e1002629. https://doi.org/10.1371/journal.pgen.1002629
Bell CG, Lowe R, Adams PD, Baccarelli AA, Beck S, Bell JT, Christensen BC, Gladyshev VN, Heijmans BT, Horvath S, Ideker T, Issa JJ, Kelsey KT, Marioni RE, Reik W, Relton CL, Schalkwyk LC, Teschendorff AE, Wagner W, Zhang K, Rakyan VK (2019) DNA methylation aging clocks: challenges and recommendations. Genome Biol 20(1):249. https://doi.org/10.1186/s13059-019-1824-y
Bestor TH, Hellewell SB, Ingram VM (1984) Differentiation of two mouse cell lines is associated with hypomethylation of their genomes. Mol Cell Biol 4(9):1800–1806. https://doi.org/10.1128/mcb.4.9.1800-1806.1984
Bibikova M, Le J, Barnes B, Saedinia-Melnyk S, Zhou L, Shen R, Gunderson KL (2009) Genome-wide DNA methylation profiling using Infinium® assay. Epigenomics 1(1):177–200. https://doi.org/10.2217/epi.09.14
Bibikova M, Barnes B, Tsan C, Ho V, Klotzle B, Le JM, Delano D, Zhang L, Schroth GP, Gunderson KL, Fan JB, Shen R (2011) High density DNA methylation array with single CpG site resolution. Genomics 98(4):288–295. https://doi.org/10.1016/j.ygeno.2011.07.007
Bickle TA, Krüger DH (1993) Biology of DNA restriction. Microbiol Rev 57(2):434–450
Bird AP (1986) CpG-rich islands and the function of DNA methylation. Nature 321(6067):209–213. https://doi.org/10.1038/321209a0
Bjornsson HT, Daniele Fallin M, Feinberg AP (2004) An integrated epigenetic and genetic approach to common human disease. Trends Genet 20(8):350–358. https://doi.org/10.1016/j.tig.2004.06.009
Bocklandt S, Lin W, Sehl ME, Sánchez FJ, Sinsheimer JS, Horvath S, Vilain E (2011) Epigenetic predictor of age. PLoS ONE 6(6):e14821. https://doi.org/10.1371/journal.pone.0014821
Boks MP, Derks EM, Weisenberger DJ, Strengman E, Janson E, Sommer IE, Kahn RS, Ophoff RA (2009) The relationship of DNA methylation with age, gender and genotype in twins and healthy controls. PLOS ONE 4(8):e6767. https://doi.org/10.1371/journal.pone.0006767
Boland MJ, Nazor KL, Loring JF (2014) Epigenetic regulation of pluripotency and differentiation. Circ Res 115(2):311–324. https://doi.org/10.1161/CIRCRESAHA.115.301517
Bollati V, Schwartz J, Wright R, Litonjua A, Tarantini L, Suh H, Sparrow D, Vokonas P, Baccarelli A (2009) Decline in genomic DNA methylation through aging in a cohort of elderly subjects. Mech Ageing Dev 130(4):234–239. https://doi.org/10.1016/j.mad.2008.12.003
Ca C (1993) Are somatic cells inherently deficient in methylation metabolism? A proposed mechanism for DNA methylation loss, senescence and aging. Growth Dev Aging 57(4):261–273
Candore G, Balistreri CR, Listì F, Grimaldi MP, Vasto S, Colonna-Romano G, Franceschi C, Lio D, Caselli G, Caruso C (2006) Immunogenetics, gender, and longevity. Ann N Y Acad Sci 1089:516–537. https://doi.org/10.1196/annals.1386.051
Cao K, Feng Z, Gao F, Zang W, Liu J (2021) Mitoepigenetics: an intriguing regulatory layer in aging and metabolic-related diseases. Free Radic Biol Med 177:337–346. https://doi.org/10.1016/j.freeradbiomed.2021.10.031
Carja O, MacIsaac JL, Mah SM, Henn BM, Kobor MS, Feldman MW, Fraser HB (2017) Worldwide patterns of human epigenetic variation. Nat Ecol Evol 1(10):1577–1583. https://doi.org/10.1038/s41559-017-0299-z
Cevenini E, Invidia L, Lescai F, Salvioli S, Tieri P, Castellani G, Franceschi C (2008) Human models of aging and longevity. Expert Opin Biol Ther 8(9):1393–1405. https://doi.org/10.1517/14712598.8.9.1393
Chen Z, Riggs AD (2011) DNA methylation and demethylation in mammals. J Biol Chem 286(21):18347–18353. https://doi.org/10.1074/jbc.R110.205286
Cho S, Ge J, Seo SB, Kim K, Lee HY, Lee SD (2014) Age estimation via quantification of signal-joint T cell receptor excision circles in Koreans. Leg Med 16(3):135–138. https://doi.org/10.1016/j.legalmed.2014.01.009
Cho S, Jung SE, Hong SR, Lee EH, Lee JH, Lee SD, Lee HY (2017) Independent validation of DNA-based approaches for age prediction in blood. Forensic Sci Int Genet 29:250–256. https://doi.org/10.1016/j.fsigen.2017.04.020
Christensen BC, Houseman EA, Marsit CJ, Zheng S, Wrensch MR, Wiemels JL, Nelson HH, Karagas MR, Padbury JF, Bueno R, Sugarbaker DJ, Yeh RF, Wiencke JK, al Kelsey KT (2009) Aging and environmental exposures alter tissue-specific DNA methylation dependent upon CpG island context. PLOS Genet 5(8):e1000602. https://doi.org/10.1371/journal.pgen.1000602
Correia Dias H, Cordeiro C, Corte Real F, Cunha E, Manco L (2020a) Age estimation based on DNA methylation using blood samples from deceased individuals. J Forensic Sci 65(2):465–470. https://doi.org/10.1111/1556-4029.14185
Correia Dias H, Cunha E, Corte Real F, Manco L (2020b) Age prediction in living: Forensic epigenetic age estimation based on blood samples. Leg Med 47:101763. https://doi.org/10.1016/j.legalmed.2020.101763
Correia Dias H, Manco L, Corte Real F, Cunha E (2021) A blood–bone–tooth model for age prediction in forensic contexts. Biology 10(10):1312. https://doi.org/10.3390/biology10121312
Cross SH, Bird AP (1995) CpG islands and genes. Curr Opin Genet Dev 5(3):309–314. https://doi.org/10.1016/0959-437X(95)80044-1
Daunay A, Baudrin LG, Deleuze J-F, How-Kit A (2019) Evaluation of six blood-based age prediction models using DNA methylation analysis by pyrosequencing. Sci Rep 9(1):8862. https://doi.org/10.1038/s41598-019-45197-w
Day K, Waite LL, Thalacker-Mercer A, West A, Bamman MM, Brooks JD, Myers RM, Absher D (2013) Differential DNA methylation with age displays both common and dynamic features across human tissues that are influenced by CpG landscape. Genome Biol 14(9):R102. https://doi.org/10.1186/gb-2013-14-9-r102
Devall M, Soanes DM, Smith AR, Dempster EL, Smith RG, Burrage J, Iatrou A, Hannon E, Troakes C, Moore K, O’Neill P, Al-Sarraj S, Schalkwyk L, Mill J, Weedon M, Lunnon K (2023) Genome-wide characterization of mitochondrial DNA methylation in human brain. Front Endocrinol 13:1059120. https://doi.org/10.3389/fendo.2022.1059120
Ehrlich M, Gama-Sosa MA, Huang LH, Midgett RM, Kuo KC, McCune RA, Gehrke C (1982) Amount and distribution of 5-methylcytosine in human DNA from different types of tissues or cells. Nucleic Acids Res 10(8):2709–2721. https://doi.org/10.1093/nar/10.8.2709
Eipel M, Mayer F, Arent T, Ferreira MR, Birkhofer C, Gerstenmaier U, Costa IG, Ritz-Timme S, Wagner W (2016) Epigenetic age predictions based on buccal swabs are more precise in combination with cell type-specific DNA methylation signatures. Aging 8(5):1034–1044. https://doi.org/10.18632/aging.100972
Ellegren H (2004) Microsatellites: simple sequences with complex evolution. Nat Rev Genet 5(6):435–445. https://doi.org/10.1038/nrg1348
El-Maarri O, Becker T, Junen J, Manzoor SS, Diaz-Lacava A, Schwaab R, Wienker T, Oldenburg J (2007) Gender specific differences in levels of DNA methylation at selected loci from human total blood: a tendency toward higher methylation levels in males. Hum Genet 122(5):505–514. https://doi.org/10.1007/s00439-007-0430-3
Feinberg AP (2001) Methylation meets genomics. Nat Genet 27(1):9–10. https://doi.org/10.1038/83825
Feng L, Peng F, Li S, Jiang L, Sun H, Ji A, Zeng C, Li C, Liu F (2018) Systematic feature selection improves accuracy of methylation-based forensic age estimation in Han Chinese males. Forensic Sci Int Genet 35:38–45. https://doi.org/10.1016/j.fsigen.2018.03.009
Field AE, Robertson NA, Wang T, Havas A, Ideker T, Adams PD (2018) DNA methylation clocks in aging: categories, causes, and consequences. Mol Cell 71(6):882–895. https://doi.org/10.1016/j.molcel.2018.08.008
Fleckhaus J, Freire-Aradas A, Rothschild MA, Schneider PM (2017) Impact of genetic ancestry on chronological age prediction using DNA methylation analysis. Forensic Sci Int Genet Suppl Ser 6:e399–e400. https://doi.org/10.1016/j.fsigss.2017.09.162
Fleischer JG, Schulte R, Tsai HH, Tyagi S, Ibarra A, Shokhirev MN, Huang L, Hetzer MW, Navlakha S (2018) Predicting age from the transcriptome of human dermal fibroblasts. Genome Biol 19(1):221. https://doi.org/10.1186/s13059-018-1599-6
Florath I, Butterbach K, Müller H, Bewerunge-Hudler M, Brenner H (2014) Cross-sectional and longitudinal changes in DNA methylation with age: an epigenome-wide analysis revealing over 60 novel age-associated CpG sites. Hum Mol Genet 23(5):1186–1201. https://doi.org/10.1093/hmg/ddt531
Fraga MF, Ballestar E, Paz MF, Ropero S, Setien F, Ballestar ML, Heine-Suñer D, Cigudosa JC, Urioste M, Benitez J, Boix-Chornet M, Sanchez-Aguilera A, Ling C, Carlsson E, Poulsen P, Vaag A, Stephan Z, Spector TD, Wu YZ, Plass C, Esteller M (2005) Epigenetic differences arise during the lifetime of monozygotic twins. Proc Natl Acad Sci U S A 102(30):10604–10609. https://doi.org/10.1073/pnas.0500398102
Fraser HB, Lam LL, Neumann SM, Kobor MS (2012) Population-specificity of human DNA methylation. Genome Biol 13(2):R8. https://doi.org/10.1186/gb-2012-13-2-r8
Freire-Aradas A, Phillips C, Mosquera-Miguel A, Girón-Santamaría L, Gómez-Tato A, Casares de Cal M, Álvarez-Dios J, Ansede-Bermejo J, Torres-Español M, Schneider PM, Pośpiech E, Branicki W, Carracedo Á, Lareu MV (2016) Development of a methylation marker set for forensic age estimation using analysis of public methylation data and the Agena Bioscience EpiTYPER system. Forensic Sci Int Genet 24:65–74. https://doi.org/10.1016/j.fsigen.2016.06.005
Freire-Aradas A, Pośpiech E, Aliferi A, Girón-Santamaría L, Mosquera-Miguel A, Pisarek A, Ambroa-Conde A, Phillips C, Casares de Cal MA, Gómez-Tato A, Spólnicka M, Woźniak A, Álvarez-Dios J, Ballard D, Court DS, Branicki W, Carracedo Á, Lareu MV (2020) A comparison of forensic age prediction models using data from four DNA methylation technologies. Front Genet 11:932. https://doi.org/10.3389/fgene.2020.00932
Freire-Aradas A, Girón-Santamaría L, Mosquera-Miguel A, Ambroa-Conde A, Phillips C, Casaresde Cal M, Gómez-Tato A, Álvarez-Dios J, Pospiech E, Aliferi A, Syndercombe Court D, Branicki W, Lareu MV (2022) A common epigenetic clock from childhood to old age. Forensic Sci Int Genet 60:102743. https://doi.org/10.1016/j.fsigen.2022.102743
Frommer M, McDonald LE, Millar DS, Collis CM, Watt F, Grigg GW, Molloy PL, Paul CL (1992) A genomic sequencing protocol that yields a positive display of 5-methylcytosine residues in individual DNA strands. Proc Natl Acad Sci 89(5):1827–1831. https://doi.org/10.1073/pnas.89.5.1827
Frumkin D, Wasserstrom A, Davidson A, Grafit A (2010) Authentication of forensic DNA samples. Forensic Sci Int Genet 4(2):95–103. https://doi.org/10.1016/j.fsigen.2009.06.009
Fuke C, Shimabukuro M, Petronis A, Sugimoto J, Oda T, Miura K, Miyazaki T, Ogura C, Okazaki Y, Jinno Y (2004) Age related changes in 5-methylcytosine content in human peripheral leukocytes and placentas: an HPLC-based study. Ann Hum Genet 68:196–204. https://doi.org/10.1046/j.1529-8817.2004.00081.x
Garagnani P, Bacalini MG, Pirazzini C, Gori D, Giuliani C, Mari D, Di Blasio AM, Gentilini D, Vitale G, Collino S, Rezzi S, Castellani G, Capri M, Salvioli S, Franceschi C (2012) Methylation of ELOVL2 gene as a new epigenetic marker of age. Aging Cell 11(6):1132–1134. https://doi.org/10.1111/acel.12005
Giuliani C, Cilli E, Bacalini MG, Pirazzini C, Sazzini M, Gruppioni G, Franceschi C, Garagnani P, Luiselli D (2016) Inferring chronological age from DNA methylation patterns of human teeth. Am J Phys Anthropol 159(4):585–595. https://doi.org/10.1002/ajpa.22921
Goldberg AD, Allis CD, Bernstein E (2007) Epigenetics: a landscape takes shape. Cell 128(4):635–638. https://doi.org/10.1016/j.cell.2007.02.006
Gopalan S, Carja O, Fagny M, Patin E, Myrick JW, McEwen LM, Mah SM, Kobor MS, Froment A, Feldman MW, Quintana-Murci L, Henn BM (2017) Trends in DNA methylation with age replicate across diverse human populations. Genetics 206(3):1659–1674. https://doi.org/10.1534/genetics.116.195594
Gopalan S, Gaige J, Henn BM (2019) DNA methylation-based forensic age estimation in human bone. bioRxiv. https://doi.org/10.1101/801647
Gršković B, Zrnec D, Vicković S, Popović M, Mršić G (2013) DNA methylation: the future of crime scene investigation? Mol Biol Rep 40(7):4349–4360. https://doi.org/10.1007/s11033-013-2525-3
Gunn PP, Walsh SJP, Roux CP (2014) The nucleic acid revolution continues—will forensic biology become forensic molecular biology? Stat Genet Methodol 5:44. https://doi.org/10.3389/fgene.2014.00044
Hamano Y, Manabe S, Morimoto C, Fujimoto S, Ozeki M, Tamaki K (2016) Forensic age prediction for dead or living samples by use of methylation-sensitive high resolution melting. Leg Med (tokyo) 21:5–10. https://doi.org/10.1016/j.legalmed.2016.05.001
Hamano Y, Manabe S, Morimoto C, Fujimoto S, Tamaki K (2017) Forensic age prediction for saliva samples using methylation-sensitive high resolution melting: exploratory application for cigarette butts. Sci Rep 7(1):10444. https://doi.org/10.1038/s41598-017-10752-w
Han Y, Franzen J, Stiehl T, Gobs M, Kuo CC, Nikolić M, Hapala J, Koop BE, Strathmann K, Ritz-Timme S, Wagner W (2020) New targeted approaches for epigenetic age predictions. BMC Biol 18(1):71. https://doi.org/10.1186/s12915-020-00807-2
Han X, Xiao C, Yi S, Li Y, Chen M, Huang D (2022) Accurate age estimation from blood samples of Han Chinese individuals using eight high-performance age-related CpG sites. Int J Legal Med 136(6):1655–1665. https://doi.org/10.1007/s00414-022-02865-3
Handelman GS, Kok HK, Chandra RV, Razavi AH, Huang S, Brooks M, Lee MJ, Asadi H (2019) Peering into the black box of artificial intelligence: evaluation metrics of machine learning methods. Am J Roentgenol 212(1):38–43. https://doi.org/10.2214/AJR.18.20224
Hannon E, Knox O, Sugden K, Burrage J, Wong CCY, Belsky DW, Corcoran DL, Arseneault L, Moffitt TE, Caspi A, Mill J (2018) Characterizing genetic and environmental influences on variable DNA methylation using monozygotic and dizygotic twins. PLoS Genet 14(8):e1007544. https://doi.org/10.1371/journal.pgen.1007544
Hannum G, Guinney J, Zhao L, Zhang L, Hughes G, Sadda S, Klotzle B, Bibikova M, Fan JB, Gao Y, Deconde R, Chen M, Rajapakse I, Friend S, Ideker T, Zhang K (2013) Genome-wide methylation profiles reveal quantitative views of human aging rates. Mol Cell 49(2):359–367. https://doi.org/10.1016/j.molcel.2012.10.016
Hao T, Guo J, Liu J, Wang J, Liu Z, Cheng X, Li J, Ren J, Li Z, Yan J, Zhang G (2021) Predicting human age by detecting DNA methylation status in hair. Electrophoresis 42(11):1255–1261. https://doi.org/10.1002/elps.202000349
Harrison A, Parle-McDermott A (2011) DNA methylation: a timeline of methods and applications. Front Genet 2:74. https://doi.org/10.3389/fgene.2011.00074
Hayatsu H (1976) Bisulfite modification of nucleic acids and their constituents. Prog Nucleic Acid Res Mol Biol 16:75–124. https://doi.org/10.1016/s0079-6603(08)60756-4
Hayatsu H, Wataya Y, Kai K, Iida S (1970) Reaction of sodium bisulfite with uracil, cytosine, and their derivatives. Biochemistry 9(14):2858–2865. https://doi.org/10.1021/bi00816a016
Heidegger A, Xavier C, Niederstätter H, de la Puente M, Pośpiech E, Pisarek A, Kayser M, Branicki W, Parson W, VISAGE Consortium (2020) Development and optimization of the VISAGE basic prototype tool for forensic age estimation. Forensic Sci Int Genet 48:102322. https://doi.org/10.1016/j.fsigen.2020.102322
Heidegger A, Pisarek A, de la Puente M, Niederstätter H, Pośpiech E, Woźniak A, Schury N, Unterländer M, Sidstedt M, Junker K, Ventayol Garcia M, Laurent FX, Ulus A, Vannier J, Bastisch I, Hedman J, Sijen T, Branicki W, Xavier C, Parson W, VISAGE Consortium (2022) Development and inter-laboratory validation of the VISAGE enhanced tool for age estimation from semen using quantitative DNA methylation analysis. Forensic Sci Int Genet 56:102596. https://doi.org/10.1016/j.fsigen.2021.102596
Heyn H, Moran S, Hernando-Herraez I, Sayols S, Gomez A, Sandoval J, Monk D, Hata K, Marques-Bonet T, Wang L, Esteller M, DNA methylation contributes to natural human variation (2013) DNA methylation contributes to natural human variation. Genome Res 23(9):1363–1372. https://doi.org/10.1101/gr.154187.112
Hill PWS, Amouroux R, Hajkova P (2014) DNA demethylation, Tet proteins and 5-hydroxymethylcytosine in epigenetic reprogramming: an emerging complex story. Genomics 104(5):324–333. https://doi.org/10.1016/j.ygeno.2014.08.012
Holländer O, Schwender K, Böhme P et al (2021) Forensische DNA-Methylierungsanalyse-Erster, technischer Ringversuch zur forensischen DNA-Methylierungsanalyse der Arbeitsgruppe „Molekulare Altersschätzung“ der Deutschen Gesellschaft für Rechtsmedizin. Rechtsmedizin 31:192–201. https://doi.org/10.1007/s00194-021-00492-7
Holliday R, Pugh JE (1975) DNA modification mechanisms and gene activity during development. Science 187:226–232. https://doi.org/10.1126/science.187.4173.226
Holmes EE, Jung M, Meller S, Leisse A, Sailer V, Zech J, Mengdehl M, Garbe LA, Uhl B, Kristiansen G, Dietrich D (2014) Performance evaluation of kits for bisulfite-conversion of DNA from tissues, cell lines, FFPE tissues, aspirates, lavages, effusions, plasma, serum, and urine. PLoS ONE 9(4):e93933. https://doi.org/10.1371/journal.pone.0093933
Hong SR, Shin K-J (2021) Bisulfite-converted DNA quantity evaluation: a multiplex quantitative real-time PCR system for evaluation of bisulfite conversion. Front Genet 12:173. https://doi.org/10.3389/fgene.2021.618955
Hong SR, Jung SE, Lee EH, Shin KJ, Yang WI, Lee HY (2017) DNA methylation-based age prediction from saliva: High age predictability by combination of 7 CpG markers. Forensic Sci Int Genet 29:118–125. https://doi.org/10.1016/j.fsigen.2017.04.006
Hong SR, Shin KJ, Jung SE, Lee EH, Lee HY (2019) Platform-independent models for age prediction using DNA methylation data. Forensic Sci Int Genet 38:39–47. https://doi.org/10.1016/j.fsigen.2018.10.005
Horvath S (2013) DNA methylation age of human tissues and cell types. Genome Biol 14:3156. https://doi.org/10.1186/gb-2013-14-10-r115
Horvath S, Raj K (2018) DNA methylation-based biomarkers and the epigenetic clock theory of ageing. Nat Rev Genet 19(6):371–384. https://doi.org/10.1038/s41576-018-0004-3
Horvath S, Zoller JA, Haghani A, Jasinska AJ, Raj K, Breeze CE, Ernst J, Vaughan KL, Mattison JA (2021) Epigenetic clock and methylation studies in the rhesus macaque. GeroScience 43(5):2441–2453. https://doi.org/10.1007/s11357-021-00429-8
Horvath S, Haghani A, Peng S, Hales EN, Zoller JA, Raj K, Larison B, Robeck TR, Petersen JL, Bellone RR, Finno CJ (2022a) DNA methylation aging and transcriptomic studies in horses. Nat Commun 13(1):40. https://doi.org/10.1038/s41467-021-27754-y
Horvath S, Lu AT, Haghani A, Zoller JA, Li CZ, Lim AR, Brooke RT, Raj K, Serres-Armero A, Dreger DL, Hogan AN, Plassais J, Ostrander EA (2022b) DNA methylation clocks for dogs and humans. Proc Natl Acad Sci 119(21):e2120887119. https://doi.org/10.1073/pnas.2120887119
Houseman EA, Accomando WP, Koestler DC, Christensen BC, Marsit CJ, Nelson HH, Wiencke JK, Kelsey KT (2012) DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics 13:86. https://doi.org/10.1186/1471-2105-13-86
Huang TH-M, Perry MR, Laux DE (1999) Methylation profiling of CpG islands in human breast cancer cells. Hum Mol Genet 8(3):459–470. https://doi.org/10.1093/hmg/8.3.459
Huang Y, Yan J, Hou J, Fu X, Li L, Hou Y (2015) Developing a DNA methylation assay for human age prediction in blood and bloodstain. Forensic Sci Int Genet 17:129–136. https://doi.org/10.1016/j.fsigen.2015.05.007
Huang CH, Chang MC, Lai YC, Lin CY, Hsu CH, Tseng BY, Hsiao CK, Lu TP, Yu SL, Hsieh ST, Chen WJ (2022) Mitochondrial DNA methylation profiling of the human prefrontal cortex and nucleus accumbens: correlations with aging and drug use. Clin Epigenetics 14(1):79. https://doi.org/10.1186/s13148-022-01300-z
Issa J-P (2014) Aging and epigenetic drift: a vicious cycle. J Clin Invest 124(1):24–29. https://doi.org/10.1172/JCI69735
Issa JP, Ottaviano YL, Celano P, Hamilton SR, Davidson NE, Baylin SB (1994) Methylation of the oestrogen receptor CpG island links ageing and neoplasia in human colon. Nat Genet 7(4):536–540. https://doi.org/10.1038/ng0894-536
Issa JP, Vertino PM, Boehm CD, Newsham IF, Baylin SB (1996) Switch from monoallelic to biallelic human IGF2 promoter methylation during aging and carcinogenesis. Proc Natl Acad Sci 93(21):11757–11762. https://doi.org/10.1073/pnas.93.21.11757
Ito H, Udono T, Hirata S, Inoue-Murayama M (2018) Estimation of chimpanzee age based on DNA methylation. Sci Rep 8(1):9998. https://doi.org/10.1038/s41598-018-28318-9
Jaffe AE, Irizarry RA (2014) Accounting for cellular heterogeneity is critical in epigenome-wide association studies. Genome Biol 15(2):R31. https://doi.org/10.1186/gb-2014-15-2-r31
Jeffreys AJ, Wilson V, Thein SL (1985a) Hypervariable ‘minisatellite’ regions in human DNA. Nature 314:67–73. https://doi.org/10.1038/314067a0
Jeffreys AJ, Wilson V, Thein SL (1985b) Individual-specific ‘fingerprints’ of human DNA. Nature 316:76–79. https://doi.org/10.1038/316076a0
Jenkins TG, Aston KI, Cairns B, Smith A, Carrell DT (2018) Paternal germ line aging: DNA methylation age prediction from human sperm. BMC Genomics 19(1):763. https://doi.org/10.1186/s12864-018-5153-4
Jiang L, Zhang K, Wei X, Li J, Wang S, Wang Z, Zhou Y, Zha L, Luo H, Song F (2023) Developing a male-specific age predictive model based on Y-CpGs for forensic analysis. Forensic Sci Int 343:111566. https://doi.org/10.1016/j.forsciint.2023.111566
Jobling MA, Gill P (2004) Encoded evidence: DNA in forensic analysis. Nat Rev Genet 5(10):739–751. https://doi.org/10.1038/nrg1455
Jones PA (1999) The DNA methylation paradox. Trends Genet 15(1):34–37. https://doi.org/10.1016/S0168-9525(98)01636-9
Jones PA (2012) Functions of DNA methylation: islands, start sites, gene bodies and beyond. Nat Rev Genet 13(7):484–492. https://doi.org/10.1038/nrg3230
Jones PA, Taylor SM (1980) Cellular differentiation, cytidine analogs and DNA methylation. Cell 20(1):85–93. https://doi.org/10.1016/0092-8674(80)90237-8
Jung SE, Lim SM, Hong SR, Lee EH, Shin KJ, Lee HY (2019) DNA methylation of the ELOVL2, FHL2, KLF14, C1orf132/MIR29B2C, and TRIM59 genes for age prediction from blood, saliva, and buccal swab samples. Forensic Sci Int Genet 38:1–8. https://doi.org/10.1016/j.fsigen.2018.09.010
Kader F, Ghai M (2015) DNA methylation and application in forensic sciences. Forensic Sci Int 249:255–265. https://doi.org/10.1016/j.forsciint.2015.01.037
Kader F, Ghai M, Olaniran AO (2020) Characterization of DNA methylation-based markers for human body fluid identification in forensics: a critical review. Int J Legal Med 134(1):1–20. https://doi.org/10.1007/s00414-019-02181-3
Kai K, Tsuruo T, Hayatsu H (1974) The effect of bisulfite modification on the template activity of DNA for DNA polymerase I. Nucleic Acids Res 1(7):889–900. https://doi.org/10.1093/nar/1.7.889
Kananen L, Marttila S (2021) Ageing-associated changes in DNA methylation in X and Y chromosomes. Epigenetics Chromatin 14(1):33. https://doi.org/10.1186/s13072-021-00407-6
Karlsson AO, Svensson A, Marklund A, Holmlund G (2008) Estimating human age in forensic samples by analysis of telomere repeats. Forensic Sci Int Genet Suppl Ser 1:569–571. https://doi.org/10.1016/j.fsigss.2007.10.153
Kaur G, Begum R, Thota S, Batra S (2019) A systematic review of smoking-related epigenetic alterations. Arch Toxicol 93(10):2715–2740. https://doi.org/10.1007/s00204-019-02562-y
Koch CM, Wagner W (2011) Epigenetic-aging-signature to determine age in different tissues. Aging 3(10):1018–1027. https://doi.org/10.18632/aging.100395
Kondo M, Aboshi H, Yoshikawa M, Ogata A, Murayama R, Takei M, Aizawa S (2021) A newly developed age estimation method based on CpG methylation of teeth-derived DNA using real-time methylation-specific PCR. J Oral Sci 63(1):54–58. https://doi.org/10.2334/josnusd.20-0138
Koop BE, Mayer F, Gündüz T, Blum J, Becker J, Schaffrath J, Wagner W, Han Y, Boehme P, Ritz-Timme S (2021) Postmortem age estimation via DNA methylation analysis in buccal swabs from corpses in different stages of decomposition—a “proof of principle” study. Int J Legal Med 135(1):167–173. https://doi.org/10.1007/s00414-020-02360-7
Laurent L, Wong E, Li G, Huynh T, Tsirigos A, Ong CT, Low HM, Kin Sung KW, Rigoutsos I, Loring J, Wei CL (2010) Dynamic changes in the human methylome during differentiation. Genome Res 20(3):320–331. https://doi.org/10.1101/gr.101907.109
Lee K, Pausova Z (2013) Cigarette smoking and DNA methylation. Front Genet 4:132. https://doi.org/10.3389/fgene.2013.00132
Lee H-C, Pang C-Y, Hsu H-S, Wei Y-H (1994) Differential accumulations of 4,977 bp deletion in mitochondrial DNA of various tissues in human ageing. Biochim Biophys Acta BBA Mol Basis Dis 1226(1):37–43. https://doi.org/10.1016/0925-4439(94)90056-6
Lee HY, Jung SE, Oh YN, Choi A, Yang WI, al Shin KJ (2015) Epigenetic age signatures in the forensically relevant body fluid of semen: a preliminary study. Forensic Sci Int Genet 19:28–34. https://doi.org/10.1016/j.fsigen.2015.05.014
Lee JW, Choung CM, Jung JY, Lee HY, Lim SK (2018) A validation study of DNA methylation-based age prediction using semen in forensic casework samples. Leg Med 31:74–77. https://doi.org/10.1016/j.legalmed.2018.01.005
Lee HY, Hong SR, Lee JE, Hwang IK, Kim NY, Lee JM, Fleckhaus J, Jung SE, Lee YH (2020) Epigenetic age signatures in bones. Forensic Sci Int Genet 46:102261. https://doi.org/10.1016/j.fsigen.2020.102261
Lee JE, Lee JM, Naue J et al (2022) A collaborative exercise on DNA methylation-based age prediction and body fluid typing. Forensic Sci Int Genet 57:102656. https://doi.org/10.1016/j.fsigen.2021.102656
Lemesh VA, Kipen VN, Bahdanava MV, Burakova AA, Bulgak AG, Bayda AV, Bruskin SA, Zotova OV, Dobysh OI (2021) Determination of human chronological age from biological samples based on the analysis of methylation of CpG dinucleotides. Russ J Genet 57:1389–1397. https://doi.org/10.1134/S1022795421120097
Levine ME, Lu AT, Quach A, Chen BH, Assimes TL, Bandinelli S, Hou L, Baccarelli AA, Stewart JD, Li Y, Whitsel EA, Wilson JG, Reiner AP, Aviv A, Lohman K, Liu Y, Ferrucci L, Horvath S (2018) An epigenetic biomarker of aging for lifespan and healthspan. Aging 10(4):573–591. https://doi.org/10.18632/aging.101414
Li E, Beard C, Jaenisch R (1993) Role for DNA methylation in genomic imprinting. Nature 366:362–365. https://doi.org/10.1038/366362a0
Li L, Song F, Huang Y, Zhu H, Hou Y (2017) Age-associated DNA methylation determination of semen by pyrosequencing in Chinese Han population. Forensic Sci Int Genet Suppl Ser 6:e99–e100. https://doi.org/10.1016/j.fsigss.2017.09.042
Li L, Song F, Lang M, Hou J, Wang Z, Prinz M, Hou Y (2020a) Methylation-based age prediction using pyrosequencing platform from seminal stains in Han Chinese males. J Forensic Sci 65(2):610–619. https://doi.org/10.1111/1556-4029.14186
Li S, Lund JB, Christensen K, Baumbach J, Mengel-From J, Kruse T, Li W, Mohammadnejad A, Pattie A, Marioni RE, Deary IJ, Tan Q (2020b) Exploratory analysis of age and sex dependent DNA methylation patterns on the X-chromosome in whole blood samples. Genome Med 12(1):39. https://doi.org/10.1186/s13073-020-00736-3
Lister R, Pelizzola M, Dowen RH, Hawkins RD, Hon G, Tonti-Filippini J, Nery JR, Lee L, Ye Z, Ngo QM, Edsall L, Antosiewicz-Bourget J, Stewart R, Ruotti V, Millar AH, Thomson JA, Ren B, Ecker JR (2009) Human DNA methylomes at base resolution show widespread epigenomic differences. Nature 462(7271):315–322. https://doi.org/10.1038/nature08514
Liu B, Du Q, Chen L, Fu G, Li S, Fu L, Zhang X, Ma C, Bin C (2016) CpG methylation patterns of human mitochondrial DNA. Sci Rep. https://doi.org/10.1038/srep23421
López-Otín C, Blasco MA, Partridge L, Serrano M, Kroemer G (2013) The hallmarks of aging. Cell 153(6):1194–1217. https://doi.org/10.1016/j.cell.2013.05.039
Lu AT, Quach A, Wilson JG, Reiner AP, Aviv A, Raj K, Hou L, Baccarelli AA, Li Y, Stewart JD, Whitsel EA, Assimes TL, Ferrucci L, Horvath S (2019) DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging 11(2):303–327. https://doi.org/10.18632/AGING.101684
Lu AT, Fei Z, Haghani A, et al (2021) Universal DNA methylation age across mammalian tissues. bioRxiv 2021.01.18.426733
Lund JB, Li S, Christensen K, Mengel-From J, Soerensen M, Marioni RE, Starr J, Pattie A, Deary IJ, Baumbach J, Tan Q (2020) Age-dependent DNA methylation patterns on the Y chromosome in elderly males. Aging Cell 19(2):e12907. https://doi.org/10.1111/acel.12907
Ma DK, Jang MH, Guo JU, Kitabatake Y, Chang ML, Pow-Anpongkul N, Flavell RA, Lu B, Ming GL, Song H (2009) Neuronal activity–induced Gadd45b promotes epigenetic DNA demethylation and adult neurogenesis. Science 323(5917):1074–1077. https://doi.org/10.1126/science.1166859
Maegawa S, Hinkal G, Kim HS, Shen L, Zhang L, Zhang J, Zhang N, Liang S, Donehower LA, Issa JP (2010) Widespread and tissue specific age-related DNA methylation changes in mice. Genome Res 20(3):332–340. https://doi.org/10.1101/gr.096826.109
Marioni RE, Shah S, McRae AF, Chen BH, Colicino E, Harris SE, Gibson J, Henders AK, Redmond P, Cox SR, Pattie A, Corley J, Murphy L, Martin NG, Montgomery GW, Feinberg AP, Fallin MD, Multhaup ML, Jaffe AE, Joehanes R, Schwartz J, Just AC, Lunetta KL, Murabito JM, Starr JM, Horvath S, Baccarelli AA, Levy D, Visscher PM, Wray NR, Deary IJ (2015) DNA methylation age of blood predicts all-cause mortality in later life. Genome Biol 16(1):25. https://doi.org/10.1186/s13059-015-0584-6
Márquez-Ruiz AB, González-Herrera L, de Luna J, Valenzuela A (2020) DNA methylation levels and telomere length in human teeth: usefulness for age estimation. Int J Legal Med 134(2):451–459. https://doi.org/10.1007/s00414-019-02242-7
Martin EM, Fry RC (2018) Environmental influences on the epigenome: exposure-associated DNA methylation in human populations. Annu Rev Public Health 39:309–333. https://doi.org/10.1146/annurev-publhealth-040617-014629
Martino D, Loke YJ, Gordon L, Ollikainen M, Cruickshank MN, Saffery R, Craig JM (2013) Longitudinal, genome-scale analysis of DNA methylation in twins from birth to 18 months of age reveals rapid epigenetic change in early life and pair-specific effects of discordance. Genome Biol 14(5):R42. https://doi.org/10.1186/gb-2013-14-5-r42
Marttila S, Jylhävä J, Nevalainen T, Nykter M, Jylhä M, Hervonen A, Tserel L, Peterson P, Hurme M (2013) Transcriptional analysis reveals gender-specific changes in the aging of the human immune system. PLoS ONE 8(6):e66229. https://doi.org/10.1371/journal.pone.0066229
Mawlood SK, Dennany L, Watson N, Pickard BS (2016) The EpiTect methyl qPCR assay as novel age estimation method in forensic biology. Forensic Sci Int 264:132–138. https://doi.org/10.1016/j.forsciint.2016.03.047
Mayer W, Niveleau A, Walter J, Fundele R, Haaf T (2000) Demethylation of the zygotic paternal genome. Nature 403:501–502. https://doi.org/10.1038/35000656
McGreevy KM, Radak Z, Torma F, Jokai M, Lu AT, Belsky DW, Binder A, Marioni RE, Ferrucci L, Pośpiech E, Branicki W, Ossowski A, Sitek A, Spólnicka M, Raffield LM, Reiner AP, Cox S, Kobor M, Corcoran DL, Horvath S (2023) DNAmFitAge: biological age indicator incorporating physical fitness. Aging. https://doi.org/10.18632/aging.204538
Meissner C, Ritz-Timme S (2010) Molecular pathology and age estimation. Forensic Sci Int 203(1–3):34–43. https://doi.org/10.1016/j.forsciint.2010.07.010
Melis JP, Jonker MJ, Vijg J, Hoeijmakers JH, Breit TM, van Steeg H (2013) Aging on a different scale—chronological versus pathology-related aging. Aging 5(10):782–788
Michael BMD (1995) Advanced protein glycosylation in diabetes and aging. Annu Rev Med 46:223–234. https://doi.org/10.1146/annurev.med.46.1.223
Nakayashiki N, Takamiya M, Shimamoto K, Aoki Y (2009) Analysis of the methylation profiles in imprinted genes applicable to parental allele discrimination. Leg Med 11. Supplement 1:S471–S472. https://doi.org/10.1016/j.legalmed.2009.02.013
Naue J, Hoefsloot HCJ, Mook ORF, Rijlaarsdam-Hoekstra L, van der Zwalm MCH, Henneman P, Kloosterman AD, Verschure PJ (2017) Chronological age prediction based on DNA methylation: massive parallel sequencing and random forest regression. Forensic Sci Int Genet 31:19–28. https://doi.org/10.1016/j.fsigen.2017.07.015
Naue J, Hoefsloot HCJ, Kloosterman AD, Verschure PJ (2018a) Forensic DNA methylation profiling from minimal traces: how low can we go? Forensic Sci Int Genet 33:17–23. https://doi.org/10.1016/j.fsigen.2017.11.004
Naue J, Sänger T, Hoefsloot HCJ, Lutz-Bonengel S, Kloosterman AD, Verschure PJ (2018b) Proof of concept study of age-dependent DNA methylation markers across different tissues by massive parallel sequencing. Forensic Sci Int Genet 36:152–159. https://doi.org/10.1016/j.fsigen.2018.07.007
Naue J, Pfeifer M, Augustin C et al (2021a) Forensische DNA-Methylierungsanalyse-Zweiter, technischer Ringversuch zur forensischen DNA-Methylierungsanalyse der Arbeitsgruppe „Molekulare Altersschätzung“ der Deutschen Gesellschaft für Rechtsmedizin. Rechtsmedizin 31:202–216. https://doi.org/10.1007/s00194-021-00493-6
Naue J, Winkelmann J, Schmidt U, Lutz-Bonengel S (2021b) Analysis of age-dependent DNA methylation changes in plucked hair samples using massive parallel sequencing. Rechtsmedizin 31:226–233. https://doi.org/10.1007/s00194-021-00487-4
Oswald J, Engemann S, Lane N, Mayer W, Olek A, Fundele R, Dean W, Reik W, Walter J (2000) Active demethylation of the paternal genome in the mouse zygote. Curr Biol 10(8):475–478. https://doi.org/10.1016/S0960-9822(00)00448-6
Ou XL, Gao J, Wang H, Wang HS, Lu HL, Sun HY (2012) Predicting human age with bloodstains by sjTREC quantification. PLoS ONE 7(8):e42412. https://doi.org/10.1371/journal.pone.0042412
Pan C, Yi S, Xiao C, Huang Y, Chen X, Huang D (2020) The evaluation of seven age-related CpGs for forensic purpose in blood from Chinese Han population. Forensic Sci Int Genet 46:102251. https://doi.org/10.1016/j.fsigen.2020.102251
Park JL, Kim JH, Seo E, Bae DH, Kim SY, Lee HC, Woo KM, Kim YS (2016) Identification and evaluation of age-correlated DNA methylation markers for forensic use. Forensic Sci Int Genet 23:64–70. https://doi.org/10.1016/j.fsigen.2016.03.005
Peters MJ, Joehanes R, Pilling LC et al (2015) The transcriptional landscape of age in human peripheral blood. Nat Commun 6:8570. https://doi.org/10.1038/ncomms9570
Pfeifer M, Bajanowski T, Helmus J, Poetsch M (2020) Inter-laboratory adaption of age estimation models by DNA methylation analysis—problems and solutions. Int J Legal Med 134(3):953–961. https://doi.org/10.1007/s00414-020-02263-7
Pidsley R, Zotenko E, Peters TJ, Lawrence MG, Risbridger GP, Molloy P, Van Djik S, Muhlhausler B, Stirzaker C, Clark SJ (2016) Critical evaluation of the Illumina MethylationEPIC BeadChip microarray for whole-genome DNA methylation profiling. Genome Biol 17(1):208. https://doi.org/10.1186/s13059-016-1066-1
Piniewska-Róg D, Heidegger A, Pośpiech E, Xavier C, Pisarek A, Jarosz A, Woźniak A, Wojtas M, Phillips C, Kayser M, Parson W, Branicki W, VISAGE Consortium (2021) Impact of excessive alcohol abuse on age prediction using the VISAGE enhanced tool for epigenetic age estimation in blood. Int J Legal Med 135(6):2209–2219. https://doi.org/10.1007/s00414-021-02665-1
Pisarek A, Pośpiech E, Heidegger A, Xavier C, Papież A, Piniewska-Róg D, Kalamara V, Potabattula R, Bochenek M, Sikora-Polaczek M, Macur A, Woźniak A, Janeczko J, Phillips C, Haaf T, Polańska J, Parson W, Kayser M, Branicki W (2021) Epigenetic age prediction in semen—marker selection and model development. Aging 13(15):19145–19164. https://doi.org/10.18632/aging.203399
Poulsen P, Esteller M, Vaag A, Fraga MF (2007) The epigenetic basis of twin discordance in age-related diseases. Pediatr Res 61:38R-42R. https://doi.org/10.1203/pdr.0b013e31803c7b98
Prado NA, Brown JL, Zoller JA, Haghani A, Yao M, Bagryanova LR, Campana MG, Maldonado J, Raj K, Schmitt D, Robeck TR, Horvath S (2021) Epigenetic clock and methylation studies in elephants. Aging Cell 20(7):e13414. https://doi.org/10.1111/acel.13414
Raj K, Szladovits B, Haghani A, Zoller JA, Li CZ, Black P, Maddox D, Robeck TR, Horvath S (2021) Epigenetic clock and methylation studies in cats. GeroScience 43(5):2363–2378. https://doi.org/10.1007/s11357-021-00445-8
Rakyan VK, Down TA, Maslau S, Andrew T, Yang TP, Beyan H, Whittaker P, McCann OT, Finer S, Valdes AM, Leslie RD, Deloukas P, Spector TD (2010) Human aging-associated DNA hypermethylation occurs preferentially at bivalent chromatin domains. Genome Res 20(4):434–439. https://doi.org/10.1101/gr.103101.109
Rand AC, Jain M, Eizenga JM, Musselman-Brown A, Olsen HE, Akeson M, Paten B (2017) Mapping DNA methylation with high-throughput nanopore sequencing. Nat Methods 14(4):411–413. https://doi.org/10.1038/nmeth.4189
Rauch TA, Wu X, Zhong X, Riggs AD, Pfeifer GP (2009) A human B cell methylome at 100−base pair resolution. Proc Natl Acad Sci 106(3):671–678. https://doi.org/10.1073/pnas.0812399106
Razin A, Riggs AD (1980) DNA methylation and gene function. Science 210:604–610. https://doi.org/10.1126/science.6254144
Ren X, Kuan PF (2020) RNAAgeCalc: a multi-tissue transcriptional age calculator. PLoS ONE 15(8):e0237006. https://doi.org/10.1371/journal.pone.0237006
Reynolds CA, Tan Q, Munoz E, Jylhävä J, Hjelmborg J, Christiansen L, Hägg S, Pedersen NL (2020) A decade of epigenetic change in aging twins: genetic and environmental contributions to longitudinal DNA methylation. Aging Cell 19(8):e13197. https://doi.org/10.1111/acel.13197
Riggs AD (1975) X inactivation, differentiation, and DNA methylation. Cytogenet Cell Genet 14(1):9–25. https://doi.org/10.1159/000130315
Ritz-Timme S, Cattaneo C, Collins MJ, Waite ER, Schütz HW, Kaatsch HJ, Borrman HI (2000) Age estimation: the state of the art in relation to the specific demands of forensic practise. Int J Legal Med 113(3):129–136. https://doi.org/10.1007/s004140050283
Ritz-Timme S, Collins MJ (2002) Racemization of aspartic acid in human proteins. Ageing Res Rev 1(1):43–59. https://doi.org/10.1016/S0047-6374(01)00363-3
Schmeling A, Olze A, Reisinger W, Geserick G (2004) Forensic age diagnostics of living people undergoing criminal proceedings. Forensic Sci Int 144(2–3):243–245. https://doi.org/10.1016/j.forsciint.2004.04.059
Schmeling A, Reisinger W, Geserick G, Olze A (2006) Age estimation of unaccompanied minors: Part I. General considerations. Forensic Sci Int 159(Supplement):S61–S64. https://doi.org/10.1016/j.forsciint.2006.02.017
Schmeling A, Geserick G, Reisinger W, Olze A (2007) Age estimation. Forensic Sci Int 165(2–3):178–181. https://doi.org/10.1016/j.forsciint.2006.05.016
Schneider PM, Prainsack B, Kayser M (2019) Erweiterte Forensische DNA-Analyse Zur Vorhersage Von Aussehen Und Biogeografischer Herkunft. Dtsch Arztebl Int 116:873–880. https://doi.org/10.3238/arztebl.2019.0873
Schwender K, Holländer O, Klopfleisch S, Eveslage M, Danzer MF, Pfeiffer H, Vennemann M (2021) Development of two age estimation models for buccal swab samples based on 3 CpG sites analyzed with pyrosequencing and minisequencing. Forensic Sci Int Genet 53:102521. https://doi.org/10.1016/j.fsigen.2021.102521
Shapiro R, Braverman B, Louis JB, Servis RE (1973) Nucleic acid reactivity and conformation: II. Reaction of cytosine and uracil with sodium bisulfite. J Biol Chem 248(11):4060–4064. https://doi.org/10.1016/S0021-9258(19)43838-6
Shi L, Jiang F, Ouyang F, Zhang J, Wang Z, Shen X (2018) DNA methylation markers in combination with skeletal and dental ages to improve age estimation in children. Forensic Sci Int Genet 33:1–9. https://doi.org/10.1016/j.fsigen.2017.11.005
Sijen T, Harbison S (2021) On the identification of body fluids and tissues: a crucial link in the investigation and solution of crime. Genes 12(11):1728. https://doi.org/10.3390/genes12111728
Simpson JT, Workman RE, Zuzarte PC, David M, Dursi LJ, Timp W (2017) Detecting DNA cytosine methylation using nanopore sequencing. Nat Methods 14(4):407–410. https://doi.org/10.1038/nmeth.4184
Slieker RC, Relton CL, Gaunt TR, Slagboom PE, Heijmans BT (2018) Age-related DNA methylation changes are tissue-specific with ELOVL2 promoter methylation as exception. Epigenetics Chromatin 11(1):25. https://doi.org/10.1186/s13072-018-0191-3
Smeers I, Decorte R, de Voorde WV, Bekaert B (2018) Evaluation of three statistical prediction models for forensic age prediction based on DNA methylation. Forensic Sci Int Genet 34:128–133. https://doi.org/10.1016/j.fsigen.2018.02.008
So MH, Lee HY (2021) Genetic analyzer-dependent DNA methylation detection and its application to existing age prediction models. Electrophoresis 42(14–15):1497–1506. https://doi.org/10.1002/elps.202000312
Soares Bispo Santos Silva D, Antunes J, Balamurugan K, Duncan G, Sampaio Alho C, McCord B (2015) Evaluation of DNA methylation markers and their potential to predict human aging. Electrophoresis 36(15):1775–1780. https://doi.org/10.1002/elps.201500137
Spólnicka M, Piekarska RZ, Jaskuła E, Basak GW, Jacewicz R, Pięta A, Makowska Ż, Jedrzejczyk M, Wierzbowska A, Pluta A, Robak T, Berent J, Branicki W, Jędrzejczak W, Lange A, Płoski R (2016) Donor age and C1orf132/MIR29B2C determine age-related methylation signature of blood after allogeneic hematopoietic stem cell transplantation. Clin Epigenetics 8(1):93. https://doi.org/10.1186/s13148-016-0257-7
Spólnicka M, Pośpiech E, Adamczyk JG, Freire-Aradas A, Pepłońska B, Zbieć-Piekarska R, Makowska Ż, Pięta A, Lareu MV, Phillips C, Płoski R, Żekanowski C, Branicki W (2018a) Modified aging of elite athletes revealed by analysis of epigenetic age markers. Aging 10(2):241–252. https://doi.org/10.18632/aging.101385
Spólnicka M, Pośpiech E, Pepłońska B, Zbieć-Piekarska R, Makowska Ż, Pięta A, Karłowska-Pik J, Ziemkiewicz B, Wężyk M, Gasperowicz P, Bednarczuk T, Barcikowska M, Żekanowski C, Płoski R, Branicki W (2018b) DNA methylation in ELOVL2 and C1orf132 correctly predicted chronological age of individuals from three disease groups. Int J Legal Med 132(1):1–11. https://doi.org/10.1007/s00414-017-1636-0
Spólnicka M, Zbieć-Piekarska R, Karp M, Machnicki MM, Własiuk P, Makowska Ż, Pięta A, Gambin T, Gasperowicz P, Branicki W, Giannopoulos K, Stokłosa T, Płoski R (2018c) DNA methylation signature in blood does not predict calendar age in patients with chronic lymphocytic leukemia but may alert to the presence of disease. Forensic Sci Int Genet 34:e15–e17. https://doi.org/10.1016/j.fsigen.2018.02.004
Tahiliani M, Koh KP, Shen Y, Pastor WA, Bandukwala H, Brudno Y, Agarwal S, Iyer LM, Liu DR, Aravind L, Rao A (2009) Conversion of 5-methylcytosine to 5-hydroxymethylcytosine in mammalian DNA by MLL partner TET1. Science 324:930–935. https://doi.org/10.1126/science.1170116
Takasaki T, Tsuji A, Ikeda N, Ohishi M (2003) Age estimation in dental pulp DNA based on human telomere shortening. Int J Legal Med 117(4):232–234. https://doi.org/10.1007/s00414-003-0376-5
Teschendorff AE, Breeze CE, Zheng SC, Beck S (2017) A comparison of reference-based algorithms for correcting cell-type heterogeneity in Epigenome-Wide Association Studies. BMC Bioinformatics. https://doi.org/10.1186/s12859-017-1511-5
Teschendorff AE, Menon U, Gentry-Maharaj A, Ramus SJ, Weisenberger DJ, Shen H, Campan M, Noushmehr H, Bell CG, Maxwell AP, Savage DA, Mueller-Holzner E, Marth C, Kocjan G, Gayther SA, Jones A, Beck S, Wagner W, Laird PW, Jacobs IJ, Widschwendter M (2010) Age-dependent DNA methylation of genes that are suppressed in stem cells is a hallmark of cancer. Genome Res 20(4):440–446. https://doi.org/10.1101/gr.103606.109
Teschendorff AE, West J, Beck S (2013) Age-associated epigenetic drift: implications, and a case of epigenetic thrift? Hum Mol Genet 22(R1):R7–R15. https://doi.org/10.1093/hmg/ddt375
Theda C, Hwang SH, Czajko A, Loke YJ, Leong P, Craig JM (2018) Quantitation of the cellular content of saliva and buccal swab samples. Sci Rep 8(1):6944. https://doi.org/10.1038/s41598-018-25311-0
Thong Z, Chan XLS, Tan JYY, Loo ES, Syn CKC (2017) Evaluation of DNA methylation-based age prediction on blood. Forensic Sci Int Genet Suppl Ser 6:e249–e251. https://doi.org/10.1016/J.FSIGSS.2017.09.095
Tsuji A, Ishiko A, Takasaki T, Ikeda N (2002) Estimating age of humans based on telomere shortening. Forensic Sci Int 126(3):197–199. https://doi.org/10.1016/S0379-0738(02)00086-5
van Dongen J, Nivard MG, Willemsen G, Hottenga JJ, Helmer Q, Dolan CV, Ehli EA, Davies GE, van Iterson M, Breeze CE, Beck S, BIOS Consortium, Suchiman HE, Jansen R, van Meurs JB, Heijmans BT, Slagboom PE, Boomsma DI (2016) Genetic and environmental influences interact with age and sex in shaping the human methylome. Nat Commun 7:11115. https://doi.org/10.1038/ncomms11115
Vidaki A, Ballard D, Aliferi A, Miller TH, Barron LP, Syndercombe Court D (2017) DNA methylation-based forensic age prediction using artificial neural networks and next generation sequencing. Forensic Sci Int Genet 28:225–236. https://doi.org/10.1016/j.fsigen.2017.02.009
Vidaki A, Daniel B, Court DS (2013) Forensic DNA methylation profiling—potential opportunities and challenges. Forensic Sci Int Genet 7(5):499–507. https://doi.org/10.1016/j.fsigen.2013.05.004
Vidaki A, González DM, Jiménez BP, Kayser M (2021) Male-specific age estimation based on Y-chromosomal DNA methylation. Aging 13(5):6442–6458. https://doi.org/10.18632/aging.202775
Waddington CH (2012) The Epigenotype. 1942. Int J Epidemiol 41(1):10–13. https://doi.org/10.1093/ije/dyr184
Wallace EV, Stoddart D, Heron AJ, Mikhailova E, Maglia G, Donohoe TJ, Bayley H (2010) Identification of epigenetic DNA modifications with a protein nanopore. Chem Commun Camb Engl 46(43):8195–8197. https://doi.org/10.1039/c0cc02864a
Wang RY-H, Gehrke CW, Ehrlich M (1980) Comparison of bisulfite modification of 5-methyldeoxycytidine and deoxycytidine residues. Nucleic Acids Res 8(20):4777–4790. https://doi.org/10.1093/nar/8.20.4777
Weber M, Davies JJ, Wittig D, Oakeley EJ, Haase M, Lam WL, Schübeler D (2005) Chromosome-wide and promoter-specific analyses identify sites of differential DNA methylation in normal and transformed human cells. Nat Genet 37(8):853–862. https://doi.org/10.1038/ng1598
Weidner CI, Lin Q, Koch CM, Eisele L, Beier F, Ziegler P, Bauerschlag DO, Jöckel KH, Erbel R, Mühleisen TW, Zenke M, Brümmendorf TH, Wagner W (2014) Aging of blood can be tracked by DNA methylation changes at just three CpG sites. Genome Biol 15(2):R24. https://doi.org/10.1186/gb-2014-15-2-r24
Weidner CI, Ziegler P, Hahn M, Brümmendorf TH, Ho AD, Dreger P, Wagner W (2015) Epigenetic aging upon allogeneic transplantation: the hematopoietic niche does not affect age-associated DNA methylation. Leukemia 29(4):985–988. https://doi.org/10.1038/leu.2014.323
Wilson VL, Jones PA (1983) DNA methylation decreases in aging but not in immortal cells. Science 220:1055–1057. https://doi.org/10.1126/science.6844925
Woźniak A, Heidegger A, Piniewska-Róg D, Pośpiech E, Xavier C, Pisarek A, Kartasińska E, Boroń M, Freire-Aradas A, Wojtas M, de la Puente M, Niederstätter H, Płoski R, Spólnicka M, Kayser M, Phillips C, Parson W, Branicki W, VISAGE Consortium (2021) Development of the VISAGE enhanced tool and statistical models for epigenetic age estimation in blood, buccal cells and bones. Aging 13(5):6459–6484. https://doi.org/10.18632/aging.202783
Wu X, Zhang Y (2017) TET-mediated active DNA demethylation: mechanism, function and beyond. Nat Rev Genet 18(9):517–534. https://doi.org/10.1038/nrg.2017.33
Xu C, Qu H, Wang G, Xie B, Shi Y, Yang Y, Zhao Z, Hu L, Fang X, Yan J, Feng L (2015) A novel strategy for forensic age prediction by DNA methylation and support vector regression model. Sci Rep 5:17788. https://doi.org/10.1038/srep17788
Yang F, Qian J, Qu H, Ji Z, Li J, Hu W, Cheng F, Fang X, Yan J (2023) DNA methylation-based age prediction with bloodstains using pyrosequencing and random forest regression. Electrophoresis. https://doi.org/10.1002/elps.202200250
Zaghlool SB, Al-Shafai M, Al Muftah WA, Kumar P, Falchi M, Suhre K (2015) Association of dna methylation with age, gender, and smoking in an arab population. Clin Epigenetics 7(1):6. https://doi.org/10.1186/s13148-014-0040-6
Zahs A, Curtis BJ, Waldschmidt TJ, Brown LA, Gauthier TW, Choudhry MA, Kovacs EJ, Bird MD (2012) Alcohol and epigenetic changes: summary of the 2011 alcohol and immunology research interest group (AIRIG) meeting. Alcohol 46(8):783–787. https://doi.org/10.1016/j.alcohol.2012.05.005
Zampieri M, Ciccarone F, Calabrese R, Franceschi C, Bürkle A, Caiafa P (2015) Reconfiguration of DNA methylation in aging. Mech Ageing Dev 151:60–70. https://doi.org/10.1016/j.mad.2015.02.002
Zapico SC, Gauthier Q, Antevska A, McCord BR (2021) Identifying methylation patterns in dental pulp aging: application to age-at-death estimation in forensic anthropology. Int J Mol Sci 22(7):3717. https://doi.org/10.3390/ijms22073717
Zbieć-Piekarska R, Spólnicka M, Kupiec T, Parys-Proszek A, Makowska Ż, Pałeczka A, Kucharczyk K, Płoski R, Branicki W (2015a) Development of a forensically useful age prediction method based on DNA methylation analysis. Forensic Sci Int Genet 17:173–179. https://doi.org/10.1016/j.fsigen.2015.05.001
Zbieć-Piekarska R, Spólnicka M, Kupiec T, Makowska Ż, Spas A, Parys-Proszek A, Kucharczyk K, Płoski R, Branicki W (2015b) Examination of DNA methylation status of the ELOVL2 marker may be useful for human age prediction in forensic science. Forensic Sci Int Genet 14:161–167. https://doi.org/10.1016/j.fsigen.2014.10.002
Zhang H, Gelernter J (2017) Review: DNA methylation and alcohol use disorders: progress and challenges. Am J Addict 26(5):502–515. https://doi.org/10.1111/ajad.12465
Zhao G, Yang Q, Huang D, Yu C, Yang R, Chen H, Mei K (2005) Study on the application of parent-of-origin specific DNA methylation markers to forensic genetics. Forensic Sci Int 154(2–3):122–127. https://doi.org/10.1016/j.forsciint.2004.09.123
Ziller MJ, Gu H, Müller F, Donaghey J, Tsai LT, Kohlbacher O, De Jager PL, Rosen ED, Bennett DA, Bernstein BE, Gnirke A, Meissner A (2013) Charting a dynamic DNA methylation landscape of the human genome. Nature 500:477–481. https://doi.org/10.1038/nature12433
Ziller MJ, Müller F, Liao J, Zhang Y, Gu H, Bock C, Boyle P, Epstein CB, Bernstein BE, Lengauer T, Gnirke A, Meissner A (2011) Genomic distribution and inter-sample variation of non-CpG methylation across human cell types. PLoS Genet 7(12):e1002389. https://doi.org/10.1371/journal.pgen.1002389
Zong D, Liu X, Li J, Ouyang R, Chen P (2019) The role of cigarette smoke-induced epigenetic alterations in inflammation. Epigenetics Chromatin 12(1):65. https://doi.org/10.1186/s13072-019-0311-8
Zubakov D, Liu F, Kokmeijer I, Choi Y, van Meurs JBJ, van IJcken WFJ, Uitterlinden AG, Hofman A, Broer L, van Duijn CM, Lewin J, Kayser M (2016) Human age estimation from blood using mRNA, DNA methylation, DNA rearrangement, and telomere length. Forensic Sci Int Genet 24:33–43. https://doi.org/10.1016/j.fsigen.2016.05.014
Zubakov D, Liu F, van Zelm MC, Vermeulen J, Oostra BA, van Duijn CM, Driessen GJ, van Dongen JJ, Kayser M, Langerak AW (2010) Estimating human age from T-cell DNA rearrangements. Curr Biol 20(22):R970–R971. https://doi.org/10.1016/j.cub.2010.10.022
Acknowledgements
The author thanks Master student Lea Wörner and Ph.D. student Laura F. Schmelzer, as well as Martin Bodner (Institute of Legal Medicine, Medical University of Innsbruck) for fruitful discussions and feedback.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The author declares no conflict of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Naue, J. Getting the chronological age out of DNA: using insights of age-dependent DNA methylation for forensic DNA applications. Genes Genom 45, 1239–1261 (2023). https://doi.org/10.1007/s13258-023-01392-8
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13258-023-01392-8