Journal of Forensic Science and Medicine

ORIGINAL ARTICLE
Year
: 2019  |  Volume : 5  |  Issue : 1  |  Page : 7--12

Estimation of postmortem interval using attenuated total reflectance: Fourier transform infrared spectroscopy in adipose tissues


Haohui Zhang1, Qi Wang2, Kai Zhang1, Ruina Liu1, Shuanliang Fan1, Zhenyuan Wang1,  
1 Department of Forensic Pathology, Xi'an Jiaotong University School of Medicine, Xi'an, China
2 Department of Forensic Medicine, Chongqing Medical University, Chongqing, China

Correspondence Address:
Dr. Zhenyuan Wang
Department of Forensic Pathology, Xi'an Jiaotong University School of Medicine, Xi'an
China

Abstract

Estimation of postmortem interval (PMI) is vitally important in forensic investigations. Although many studies have examined the chemical changes of various tissues over time, no reports using spectroscopic methods in adipose tissue are available. In this study, attenuated total reflectance–Fourier transform infrared (ATR–FTIR) spectroscopy was utilized to collect comprehensive biochemical information from human adipose tissues in vitro at different times. Thereafter, mice were used as samples for in vivo experiments for more detailed studies on PMI. Then, partial least squares (PLS) model for PMI estimation was established based on the acquired spectral dataset of mouse samples. The spectral variable associated with C=O arising from lipids and free fatty acids was most susceptible to PMI. Moreover, the PLS model appeared to achieve a satisfactory prediction with a root mean square error of cross-validation of 1.78 days, and the reliability of the model was determined in an external validation set with a root mean square error of prediction of 1.87 days. The study shows the possibility of application of ATR–FTIR methods in PMI estimation using adipose tissue.



How to cite this article:
Zhang H, Wang Q, Zhang K, Liu R, Fan S, Wang Z. Estimation of postmortem interval using attenuated total reflectance: Fourier transform infrared spectroscopy in adipose tissues.J Forensic Sci Med 2019;5:7-12


How to cite this URL:
Zhang H, Wang Q, Zhang K, Liu R, Fan S, Wang Z. Estimation of postmortem interval using attenuated total reflectance: Fourier transform infrared spectroscopy in adipose tissues. J Forensic Sci Med [serial online] 2019 [cited 2019 Sep 15 ];5:7-12
Available from: http://www.jfsmonline.com/text.asp?2019/5/1/7/255131


Full Text



 Introduction



Estimation of postmortem interval (PMI), also known as time since death, is vitally important in forensic investigations. In daily forensic casework, a precise estimation of the PMI can help forensic pathologists verify the witness statement, narrow the search scope, or even guide the direction of the investigation. Various methods have been explored by forensic scholars to achieve this purpose.[1] Traditional methods include the examination of algor mortis, rigor mortis, livor mortis, and the growth state of insects after death.[2],[3] Many new methods focus on the postmortem changes in chemical and biomolecule indices, such as the degradation of DNA, RNA, protein, and the variation of postmortem microorganisms.[4],[5],[6],[7] However, most of these new methods require complex processes to analyze forensic samples, and the instruments required are expensive and time-consuming. Therefore, the use of a convenient and simple pretreatment for estimating the PMI would be quite beneficial in forensic daily work.

Fourier transform infrared (FTIR) spectroscopy is a highly sensitive analytical method for identifying the molecular composition of biological samples according to the detection of vibrational mode of their chemical bonds. Because of its nondestructive, rapid, portable, and user-friendly advantages, FTIR has been widely used in forensic investigations to detect and analyze fingerprints, inks, fibers, hair, and gunshot residues found at crime scenes.[8],[9],[10],[11],[12] In addition, FTIR can be applied to analyze biological samples, including organic macromolecule components, such as proteins, carbohydrates, lipids, and nucleic acids.[13] As organs and tissues degrade gradually after death, the components of the tissues on the remains change over time, and this phenomenon can be reflected in the spectral data, such as the decline in the peak intensities and areas. In combination with chemometrics, more spectral information can be obtained and more appropriate models can be established to estimate PMI.[14],[15]

Previous works by the authors have demonstrated the usefulness of FTIR for the study of biochemical changes in organs and biological fluids, and several characteristic peak intensities and areas were proved to be correlated with PMI.[1],[16],[17],[18],[19] Considering that most tissues are susceptible to degradation and microbial interference, there are many limitations to accurately estimating PMI for remains in a state of high putrefaction. Yan et al. found that the composition of adipocere is different at various points of time, which can be applied to estimate PMI in specific cases.[20] The degradation of lipids is relatively slow and can be applied for longer PMI estimation.[21] Moreover, adipose tissue is close to the body surface and can be obtained through a small incision, even without performing a forensic autopsy, and it may be an appropriate sample to estimate PMI of highly decomposed remains. Thus, in this work, we employed FTIR coupled with chemometrics to detect the decomposition process of adipose tissue.

 Materials and Methods



Human sample preparation

Informed consent was obtained from the relatives of all deceased individuals in this study, and it must be emphasized that all procedures in this study comply with the requirements of local laws and institutional guidelines and were approved and overseen by the Ethics Committee of Xi'an Jiaotong University. A total of eight subcutaneous adipose tissues of the abdomen were collected from eight human remains: two were highly decomposed and six were relatively fresh, as shown in [Table 1]. For each case, after a longitudinal incision of the whole abdominal wall during the autopsy, approximately 4 cm × 4 cm of skin with the full layer of subcutaneous fat was cut off from above the umbilical area. Then, the collected specimens were preserved in a 100-mL measuring cup and placed in an environmentally controlled chamber with a constant temperature of 25°C ± 1°C and a relative humidity of 50% ± 5%. Then, approximately 0.5 cm × 0.5 cm × 0.5 cm of adipose tissue was sheared from each sample every 2 days for the next 2 weeks. To obtain stable spectral images, pretreatment was performed before sample measurement: all small tubes of the adipose tissues were ultrasonically ground for 15 s and centrifuged at 4°C and 3000 rpm for 3 min, and the supernatants were obtained and then kept frozen at −80°C until FTIR analysis.{Table 1}

Animal sample preparation

Male KM mice (n = 121, weight 24–26 g), purchased from the Animal Center of Xi'an Jiaotong University, were anesthetized by 0.1 ml/10 g 4% chloral hydrate through the abdomen, and were then sacrificed by cervical dislocation. No sample was collected prior to sacrifice. All the animal experiments in the present study were specifically approved and overseen by the Care and Use of Laboratory Animal Committee of Xi'an Jiaotong University. Cadavers were kept at moderate ambient temperatures of 25°C ± 1°C and a relative humidity of 50% ± 5% in an environmentally controlled chamber following sacrifice. Adipose tissue samples from 121 mice were taken at 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, and 14 days (11 samples for each time point, eight for a calibration set, and three for a prediction set). All adipose tissue samples were collected from the inguinal region. Because the mice did not have enough lipid in their adipose tissue, all samples were ultrasonically ground for 15 s, and 100-μL petroleum ether was added to purify the lipids in adipose tissue. Moreover, it must be noted that the effumability of petroleum ether should not impact the spectral data collection. Then, the samples were centrifuged at 4°C and 3000 rpm for 3 min, and the supernatants were mainly contained in adipose tissue for FTIR analysis.

Fourier transform infrared measurements

FTIR spectra were collected by a Thermo Nicolet–IS50 FTIR spectrometer (Thermo Electron Scientific Instruments Corp., WI, USA) coupled with a diamond-attenuated total reflectance (ATR) accessory, and a KBr beam splitter was used for spectral acquisition. The infrared spectra analysis software package OMNIC version 8.2 (Thermo Nicolet Analytical Instruments, WI, USA) was used for analyzing the spectra and recording the data. Three 1-μL supernatants of each sample were added drop wise to the diamond ATR crystal. To ensure spectral reproducibility and assess analytical precision, every drop of the supernatant was detected and recorded three times. The parameters of spectral collection were set to frequencies ranging from 4000 to 400 cm−1 with a resolution of 4 cm−1 and 32 scans.

Attenuated total reflectance–Fourier transform infrared spectral pretreatment and analysis

For each sample, nine replicate spectra were obtained, and the mean spectrum was used for dataset treatment. To reduce systematic variations, we used the standard normal variate (SNV) method to normalize the spectra.[22] Partial least squares (PLS) is a popular modeling approach for high-throughput data, and its application in different fields to address a variety of problems has increased in recent years.[23] The PLS attempts to find factor latent variables (LVs) that maximize the amount of variation explained in X that is relevant for predicting response Y and modeling the linear relationship between X and Y, such as the spectral data and the PMI values in this study. A critical step in optimizing the predictive ability of the model is to determinate LVs, which usually depends on cross-validation (CV). In this case, we adopted leave-one-out CV, and the selection of an optimal number of LVs was based on the inclusion of additional factors only when the root mean square error of CV (RMSECV) improved by at least 5%. The value of the model is evaluated by RMSECV, root mean square error of prediction (RMSEP), determination coefficient of CV (Rcv2), and determination coefficient of prediction (Rp2). RMSECV generally indicates the magnitude of global model error in the calibration model. RMSEP was used to evaluate the performance of the prediction set in the prediction process. Lower RMSECV and RMSEP values, but small differences between these two values, indicate better predictive quality. The coefficient of determination (R2) can evaluate the goodness of fit between actual and predicted values, and the closer it is to one, the better the model fits. MATLAB R2017b (The MathWorks, MA, USA) equipped with PLS_Toolbox 8.6 (Eigenvector Research, Inc., 3905 West Eaglerock Drive, Wenatchee, USA) was used for data analysis.

 Results and Discussion



Visual spectral analysis of human samples

[Figure 1] shows the SNV-pretreated ATR–FTIR spectra of sample 1 at 0 and 6 days in the range of 4000–400 cm−1. The assignment of the bands is shown in [Table 2].[24],[25] The spectra have strong C–H absorption between 3050 and 2850 cm−1. The separate bands at 2922 and 2852 cm−1 correspond to asymmetrical C–H stretching (CH2) and symmetrical C–H stretching (CH2), respectively, with a weak peak at 3008 cm−1 caused by asymmetric C–H stretching (=C–H). The peak at 1743 cm−1 is due to the C =O stretching mode of lipids. Compared with the spectrum at 0 days, the spectrum at 6 days showed a new peak at 1711 cm−1, which is known as a significant region for free fatty acids. We considered that this may be a characteristic peak related to PMI. The peak at 1465 cm−1 is assigned to the CH2 scissoring mode of the acyl chain of lipids, and the frequency range of 1159–1174 cm−1 results from stretching vibrations of carbohydrates. The spectral region between 1120 and 1050 cm−1 has two weak peaks from RNA caused by C–O stretching at 1117 cm−1 and PO2− symmetric stretching at 1089 cm−1.{Figure 1}{Table 2}

The time of the peak at 1711 cm−1 first appearing (T1711) in eight samples during the 2 weeks is listed in [Table 1]. We found that T1711 of sample 1 and sample 7 are 6 and 8 days, respectively, which are far less than those of other samples, and a common characteristic of both is that they are in a state of high putrefaction with longer PMI. Unfortunately, the PMI in both cases cannot be fully determined due to the lack of on-the-spot information. Excluding the two highly decomposed cases, the T1711 of sample 2 and sample 6 are both 10 days, which are less than that in other samples. The PMI of these two samples is longer than that of the other samples. This finding further confirmed the relevance of the peak at 1711 cm−1 to PMI.

Studies have proved that the body's adipose tissue is mostly composed of lipids, of which 90%–99% are triglycerides that are hydrolyzed by intrinsic tissue lipases shortly after death to produce a mixture of saturated and unsaturated fatty acids.[21] In the study by Swann et al., fatty acids could be detected by a method of gas chromatography–mass spectrometry in fluids released during decomposition, and the components of fatty acids were correlated with the PMI.[26] We found that biomarkers (free fatty acids) can also be detected by ATR–FTIR in adipose tissue. The appearance of free fatty acids resulting from the decomposition of lipids is correlated with the PMI of the remains according to a preliminary study on human sample in vitro. However, the condition in vitro is quite different from that in vivo. Due to the lack of continuous research on lipid decomposition regularity, we studied mice to verify the findings on human samples and to establish a suitable model to estimate PMI.

Mouse sample spectral analysis

We used only the spectral regions at 3050–2800 cm−1 and 1800–400 cm−1, which contained the fundamental vibrational energy-absorbing frequencies of most biomolecules for further analysis to reduce the interference from noise. [Figure 2] shows the average spectra after pretreatment with SNV of every PMI group and their comparison. The assignments of the peaks are listed in [Table 2] and described above. Compared to the uncontrollability and limitations of human samples, we can clearly determine the postmortem trends in mouse adipose tissue. In [Figure 2], the most conspicuously changed peaks are at 1743 and 1711 cm−1, and the intensity of 1743 cm−1 decreased over time, while 1711 cm−1 increased at first and then decreased. The peak associated with = C–H at 3008 cm−1 displayed an increasing trend in intensity, while the peak related to CH2 vibration at 2922 and 2852 cm−1 displayed a trend of rising first and then falling. We believe that this change in the spectrum results from the hydrolysis of triglycerides, which reduces the content of lipids and increases the content of free fatty acids, and with an increasing PMI, free fatty acids dispersed from the adipose tissue and became part of the degradation products. Furthermore, we can determine that the time when the peak at 1711 cm−1 first appeared is on the 4th day at 25°C ± 1°C and a relative humidity of 50% ± 5%. The variation of the spectrum on the characteristic peak is consistent with the human sample in vitro. The difference is that the characteristic peak in mouse samples in vivo appeared earlier than in human samples in vitro. We consider adipose tissue in vivo to be degraded in the same way as in vitro, but happening more quickly, due to the richer content of enzymes and water. The peak at 1159–1174 cm−1, which represents carbohydrates, and the peaks related to RNA at 1117 and 1089 cm−1 decreased. This reduction in carbohydrates and ribose is consistent with that of our previous research.[17],[19] The increased intensity of the peak at 1545 cm−1, which is related to the amide II region, informed us about the mixing of short peptide chains.{Figure 2}

In the next step, we applied chemometric methods to elucidate more information from the spectral dataset to construct a model for PMI estimation. [Figure 3] shows the relatively satisfactory result (Rcv2 = 0.80, Rp2 = 0.80; RMSECV = 1.78 days, RMSEP = 1.87 days) of the PLS model with 4 LVs in leave-one-out CV after data were preprocessed by the SNV method. Furthermore, we calculated the variable importance in projection (VIP) scores for all spectral variables to account for this model.[27] Variables with VIP scores above 1.0 were considered to be influential for model construction. The higher the VIP scores for each variable, the more important the variable to the PLS model. [Figure 4] shows the highest peaks of VIP scores at 1743 and 1711 cm−1 which belonged to C=O vibrations of lipids and free fatty acids, followed by C–O vibrations from carbohydrates. Meanwhile, asymmetric stretching and scissoring of CH2 and amide II constitute less influential variables. The VIP scores further reflect that the process of triglyceride degradation into fatty acids is relatively regular and has a stable trend, which is of great significance for estimating PMI.{Figure 3}{Figure 4}

In vivo experiments in mice, we found that the postmortem changes were consistent with that of human adipose tissue in vitro after death. This finding is of great significance for the investigation of the time of death of decomposed remains in the future. Comparing the degradation rule to other tissues, changes in adipose tissue were mainly observed after 4 days, while those in the liver and spleen were before 6 days, and when the PMI goes to over 6 days, the tissues full of protein can easily be influenced by microbes according to Huang's finding.[19] In addition, Wang et al. collected spectral data of plasma and established models to estimate the PMI (within 2 days) with the application of chemometrics, which achieved satisfying results (Rcv2 = 0.91, Rp2 = 0.85, RMSECV = 4.76 h, RMSEP = 5.31 h), yielding a higher R2 than this study.[17] We found that protein-rich tissue had a better correlation with the time of death in the short term, but had limitations in inferring the PMI of degraded remains. In this study, we found that adipose tissue compensates for this limitation. As previous studies reported, some metabolites of lipids in biological samples could be considered biomarkers for PMI estimation.[20],[26],[28] However, they simply picked out possible biomarkers and did not perform any subsequent research to extrapolate the PMI. Based on their research, we established a model to estimate the PMI by studying the consistent postmortem changes of mice.

Along with the advantages of adipose tissue, the current spectroscopic study suggests that adipose tissue may be an appropriate medium for PMI estimation in a highly decomposed body. However, the PMI is affected by many factors, such as temperature, humidity, and cause of death,[29] and the current study focused on a single condition, and the application scope of the established model is limited. More complex and changeable conditions in real forensic cases might influence the process of postmortem changes of adipose tissues and lead to larger prediction errors. Therefore, in future work, we need to study the postmortem changes of adipose tissue under complex conditions, establish corresponding models, reduce errors, and adapt to the needs of real work.

 Conclusion



In this work, ATR–FTIR spectroscopy was applied to acquire biochemical information in adipose tissues of human samples in vitro for the first time, and then in vivo tests were performed on mice. The results show that the combination of FTIR analysis and chemometrics based on biochemical changes in adipose tissue is ideal for estimating the PMI in highly decomposed remains. Spectroscopic findings reflected that PMI-dependent changes in adipose tissue derived from the time-ordered process of hydrolysis of lipids into free fatty acids, and other biological molecules, such as carbohydrates and nucleic acids also have tiny effects. Moreover, the PLS model was established to predict the PMI, which produced satisfactory results. In summary, this study demonstrates the feasibility of using ATR–FTIR on adipose tissue to estimate the PMI and offers a promising new approach in the specific scene of highly decomposed remains.

Financial support and sponsorship

This study was funded by the Council of the National Natural Science Foundation of China (No. 81730056).

Conflicts of interest

There are no conflicts of interest.

References

1Li C, Wang Q, Zhang Y, Lin H, Zhang J, Huang P, et al. Research progress in the estimation of the postmortem interval by Chinese forensic scholars. Forensic Sci Res 2016;1:3-13.
2Henssge C, Madea B. Estimation of the time since death in the early post-mortem period. Forensic Sci Int 2004;144:167-75.
3Amendt J, Campobasso CP, Gaudry E, Reiter C, LeBlanc HN, Hall MJ, et al. Best practice in forensic entomology – Standards and guidelines. Int J Legal Med 2007;121:90-104.
4Frere B, Suchaud F, Bernier G, Cottin F, Vincent B, Dourel L, et al. GC-MS analysis of cuticular lipids in recent and older scavenger insect puparia. An approach to estimate the postmortem interval (PMI). Anal Bioanal Chem 2014;406:1081-8.
5Hunter MC, Pozhitkov AE, Noble PA. Accurate predictions of postmortem interval using linear regression analyses of gene meter expression data. Forensic Sci Int 2017;275:90-101.
6Li C, Ma D, Deng K, Chen Y, Huang P, Wang Z, et al. Application of MALDI-TOF MS for estimating the postmortem interval in rat muscle samples. J Forensic Sci 2017;62:1345-50.
7Metcalf JL, Xu ZZ, Weiss S, Lax S, Van Treuren W, Hyde ER, et al. Microbial community assembly and metabolic function during mammalian corpse decomposition. Science 2016;351:158-62.
8Bueno J, Sikirzhytski V, Lednev IK. Attenuated total reflectance-FT-IR spectroscopy for gunshot residue analysis: Potential for ammunition determination. Anal Chem 2013;85:7287-94.
9Crane NJ, Bartick EG, Perlman RS, Huffman S. Infrared spectroscopic imaging for noninvasive detection of latent fingerprints. J Forensic Sci 2007;52:48-53.
10Espinoza E, Przybyla J, Cox R. Analysis of fiber blends using horizontal attenuated total reflection Fourier transform infrared and discriminant analysis. Appl Spectrosc 2006;60:386-91.
11Manheim J, Doty KC, McLaughlin G, Lednev IK. Forensic hair differentiation using attenuated total reflection Fourier transform infrared (ATR FT-IR) spectroscopy. Appl Spectrosc 2016;70:1109-17.
12Zieba-Palus J, Kunicki M. Application of the micro-FTIR spectroscopy, Raman spectroscopy and XRF method examination of inks. Forensic Sci Int 2006;158:164-72.
13Olsztyńska-Janus S, Szymborska-Małek K, Gąsior-Głogowska M, Walski T, Komorowska M, Witkiewicz W, et al. Spectroscopic techniques in the study of human tissues and their components. Part I: IR spectroscopy. Acta Bioeng Biomech 2012;14:101-15.
14Byrne HJ, Knief P, Keating ME, Bonnier F. Spectral pre and post processing for infrared and Raman spectroscopy of biological tissues and cells. Chem Soc Rev 2016;45:1865-78.
15Gómez-Caravaca AM, Maggio RM, Cerretani L. Chemometric applications to assess quality and critical parameters of virgin and extra-virgin olive oil. A review. Anal Chim Acta 2016;913:1-21.
16Ping H, Wang S, Bai J, Ke Y, Fan S, Zhang P, et al. To estimate the postmortem interval using FTIR spectroscopy. Chin J Forensic Med 2011;26:104-8.
17Wang Q, He H, Li B, Lin H, Zhang Y, Zhang J, et al. UV-Vis and ATR-FTIR spectroscopic investigations of postmortem interval based on the changes in rabbit plasma. PLoS One 2017;12:e0182161.
18Zhang J, Li B, Wang Q, Wei X, Feng W, Chen Y, et al. Application of Fourier transform infrared spectroscopy with chemometrics on postmortem interval estimation based on pericardial fluids. Sci Rep 2017;7:18013.
19Huang P, Tian W, Tuo Y, Wang Z, Yang G. Estimation of postmortem interval in rat liver and spleen using Fourier transform infrared spectroscopy. Spectrosc Lett 2009;42:108-16.
20Yan F, McNally R, Kontanis EJ, Sadik OA. Preliminary quantitative investigation of postmortem adipocere formation. J Forensic Sci 2001;46:609-14.
21Dent BB, Forbes SL, Stuart BH. Review of human decomposition processes in soil. Environ Geol 2004;45:576-85.
22Dhanoa MS, Barnes RJ, Lister SJ. Standard normal variate transformation and de-trending of near-infrared diffuse reflectance spectra. Appl Spectrosc 1989;43:772-7.
23Mehmood T, Ahmed B. The diversity in the applications of partial least squares: an overview. J Chemom 2016;30:4-17.
24Man YB, Moh MH, van de Voortb. Determination of free fatty acids in crude palm oil and refined-bleached-deodorized palm olein using Fourier transform infrared spectroscopy. J Am Oil Chem Soc 1999;76:485-90.
25Movasaghi Z, Rehman S, Rehman DI. Fourier transform infrared (FTIR) spectroscopy of biological tissues. Appl Spectrosc Rev 2008;43:134-79.
26Swann L, Chidlow GE, Forbes S, Lewis SW. Preliminary studies into the characterization of chemical markers of decomposition for geoforensics. J Forensic Sci 2010;55:308-14.
27Chong IG, Jun CH. Performance of some variable selection methods when multicollinearity is present. Chemom Intell Lab Syst 2005;78:103-12.
28Wood PL, Shirley NR. Lipidomics analysis of postmortem interval: preliminary evaluation of human skeletal muscle. Metabolomics 2013;3:1-3.
29Wang Q, Zhang Y, Lin H, Zha S, Fang R, Wei X, et al. Estimation of the late postmortem interval using FTIR spectroscopy and chemometrics in human skeletal remains. Forensic Sci Int 2017;281:113-20.