Recently, the number of the unidentified cadavers is increasing due to disasters and terrorisms episodes worldwide, as well as a consequence of social internationalization 1,2.
Among the various items needed for cadaver identification, age is among the most important ones. Most of the conventional age estimation methods, including anthropological methods and dental method 3 are mainly used in highly corrupted or skeletonized corpses. Closing of cranial suture 4, change of the pubic bone 5, morphology of the ribs 6, the developmental stages of the seven left permanent mandibular teeth 7 and so on represent some examples of those methods. Some of the methods provide high estimation accuracy 8–10 However, they are mainly subjective and require highly technical skills and experience 2,11−13. Recently, Computed Tomography and Magnetic Resonance Imaging are also used in this area 14–17. However, these methods require very expensive equipment, so their use is not yet widely adopted.
Age estimation is somewhat thought to be necessary only for highly corrupted corpses, but actually many fresh corpses which requires age estimation are found on a daily basis. For age estimation in those fresh cadavers or living individuals, in addition to above mentioned methods, bone X-ray examination, physical examination, dental eruptions and so on 18, molecular biological methods are also used. For instance, among the alterations that occur to our body as we age, many relate to proteins, and attach to intrinsic variations of their molecular structure. Stadtman 19 proved earlier how the age-related increase in oxidized proteins might reflect the age-dependent accumulation of unrepaired DNA damage. The most advanced and objective concept of “age predictor” in modern forensic science matches the so-called epigenetic (or DNA, or Horvath’s) clock, 20–22 which is based on DNA methylation of CpG dinucleotides and can be used to estimate the age of a suspect based on blood remains. Usable on a wide spectrum of different tissues, the median error in age prediction of this method is 3.6 y, with a Pearson correlation coefficient, R = 0.96, to chronological age 21. The stronger the association of the two variables, the closer the Pearson correlation coefficient, r; this coefficient being either + 1 or -1 depending on whether the relationship is positive or negative, respectively. The Horvath’s clock indeed leads to a remarkably precise age estimation as compared to other proposed methods (e.g., telomere length, p16INK4a, or microsatellite mutations) 23–25, which show distinctly lower correlation coefficients and only relate to specific types of cell rather than directly to chronological age. The epigenetic clock was also found to relate to Body Mass Index (BMI) with a good correlation, R = 0.42, for analyses conducted on liver tissue 21.
However, though the Horvath’s method proved highly objective and has remarkable accuracy among the various age estimation methods, special technical skills, expensive reagents and equipment are still required as well as time and efforts.
Looking for a different approach, we examined the possibility of using Raman spectroscopy of human skin in the attempt to establish a reliable and easily accessible method of age estimation in the forensic medicine. Our spectroscopic method relies on the collection of incisional skin samples during forensic autopsy, and spectroscopic examination of structural characteristics of both proteins and lipids contained in the dermis. Aging of structures of skin has been deeply studied by employing a number of different analytical techniques 26,27. Our previously published proof-of-concept study attempted to label all the skin-emitted Raman bands and to locate any biophysical link between vibrational fingerprints and specific structural variations associated with human age 28. The existence of such links was indeed phenomenologically proved for protein structures, but it is yet hindered by insufficient statistics. The final proof for our spectroscopic findings thus relies on their statistical validations.
The main purpose here is a statistical validation of the presence of Raman spectral markers for precise age identification from human skin samples. In doing so, we used the same spectral deconvolution algorithm previously proposed 28. In that previous study, we have studied the Raman spectra of various skin samples to find a candidate of ideal structural change for age estimation with small number of cases. Accordingly, we found that protein folding may be a good potential “biological clock”. In this study, upon increasing the number of autopsy samples (i.e., n = 132) and concurrently taking care that wide age range are included, we attempted to assess the statistical exactness of age estimation by Raman spectrometry. The statistical validation presented here could be attractive for the forensic community because the Raman method is label-free and directly links vibrational features at the molecular scale to practical forensic purposes.
Figure 1 gives schematic drafts of different secondary structures of proteins.
These self-explanatory figures also include drafts of the specific vibrational modes that we monitored in this study, as well as the Raman spectroscopic links in terms of vibrational frequencies of different structural assemblies. The structure of skin upon aging has been thoroughly studied by means of numbers of different analytical techniques 26,27. Yet, explicit biophysical links are missing between the observed vibrational fingerprints and the specific structural variations associated with human age. With reference to the relative intensities of specific Raman emissions from different protein mesostructures (α-helix structure and disordered random coil protein structures in Fig. 1), we defined the protein-folding intensity ratio (RPF). RPF represents the ratio of Raman intensities related to the C = O stretching mode (i.e., in the overall Amide I vibrational spectrum depicted in Table S2) in disordered protein structures (Raman signal seen at 1681 cm-1) to α-helix structure (Raman signal at 1652 cm-1) 29–37. Some literatures 29,30 reported that proteins irreversibly fold upon aging. Accordingly in our previous study 28, the Raman features representing those folding processes were located as parameters sensing biological age, although their age-sensing precision was yet unknown to us.
The aim of this study was to statistically validate the Raman spectroscopic parameters of protein folding in the human skin as age-predicting parameters.