Can latent fingerprint disclose the sex of the donor? A preliminary test study using GC–MS analysis of latent fingerprints

While fingerprints are a highly used means of identification, not every fingerprint left behind on a potential crime scene can be used for identification purposes. In some cases, the fingerprint may be smudged, partially preserved or overlapping with other prints hence distorting the ridge pattern and may therefore be not appropriate for identification. Further, fingermark residue yields a very low abundance of genetic material for DNA analysis. In such cases, the fingermark may be used to retrieve basic donor information such as sex. The focus of this paper was to assess the possibility of differentiating between the sexes of the donor of latent fingermarks. Analytical method was GC–MS analysis of the chemical compounds of latent fingermarks using 22 male and 22 female donors. Results showed 44 identified compounds. Two alcohols, octadecanol C18 and eicosanol C20, were found to show a difference that was statistically significant between male and female donors. There is also some evidence for the possibility of distinguishing sex of the fingermark donor based on the distribution of branched chain fatty acids, as free compounds or esterified in wax esters.


| INTRODUC TI ON
The human fingerprint derives from the corrugated ridge and furrow pattern on the skin of the fingertips and is currently used as one of biometric traits for identification of individuals, even identical twins.
Fingerprints have been used for identification purposes for more than 150 years [1]. The latent fingerprint is composed of the natural secretions from the glands in the skin, mainly eccrine and sebaceous excretions [2], as well as environmental contaminants. Because the openings of the excretory glands are located at the ridges rather than the furrows, the excretion transferred from the fingertip to a surface upon touch leaves behind the distinctive fingerprint pattern.
The focus of this paper was to assess the possibility of differentiating between the sexes of the donor of latent fingermarks using GC-MS analysis of the chemical compounds of latent fingermarks.
Using this data, the difference in chemical composition of latent fingermarks is examined to provide a method for analysis of fingermarks at crime scenes that are not conducive to be lifted for identification from the visualized ridge and furrow pattern and may instead be useful for obtaining details on the donor profile.  Table 1 for donor demographics. The project received ethical approval from the University of Dundee. Donors provided informed consent in writing and their data, and all samples, were anonymized to be identified only by a number (101-122) with a pre-fix of "M" for male and "F" for female. Participants were excluded as donors if receiving steroid or hormone treatment.

| Collection of fingermarks
Donors were asked to wash their hands using only water and rinse them in distilled water. The hands were patted dry using tissue of the standard institutional brand "Lotus Professional (100% UK and Ireland recovered fibers)". Donors were requested not to touch anything and let 5 min pass for the regeneration of fingerprint residue and then requested to rub all 10 fingertips on the skin behind the ears. The fingertips of all 10 digits were then rubbed on disinfected microscope slide and placed in a glass jar. This process was repeated three times such that each donor provided three replicates, all of which were placed in the same jar for transport.

| Extraction of samples
Both sides of the slides were then dripped with dichloromethane and the extract was collected in a culture tube. The three replicates provided by the participants were all treated separately so that for each participant three samples were produced. Fifty microliters of the internal standard, methyl nonadecanoate (10.3 mg in 50 mL of dichloromethane), was added to each sample. The dichloromethane containing the fingermark residue was evaporated down by a gentle stream of nitrogen and a weak heating source to a volume of about 120 μL. The sample was then transferred to a 200 μL autosampler vial and evaporated down to approximately 60 μL. Each vial was labeled with the anonymized donor number and number of sample replicate.

| Analysis of samples by gas chromatographymass spectrometry
The samples were analyzed using a DSQ II quadrupole mass spectrometer coupled to a Trace Ultra gas chromatograph (Thermo Between each sample, a "blank" sample was inserted (pure solvent) to minimize carryover. The samples were all analyzed by the GC-MS on the same day as collection. One sample with three replicates was produced to act as controls. The data were acquired using the Xcalibur software package (V.2.07 Thermo Electron Corporation).
The mass spectra of each peak in the TIC of several representative raw data files were examined manually and specific ions characteristic of each compound in the samples were selected using Xcalibur (Table S1). Selection criteria were that the ions should be of high relative abundance and should be unique to the compound and/or be well resolved from other ions with the same m/z [39]. Summed SIC peak areas for each compound were normalized to the internal standard, by dividing the SIC area with that of the internal standard. The values quoted are therefore peak area ratios.

| Statistical analysis
MANOVA analysis was conducted to investigate the statistical differences between the mean peak area ratios of two sexes included in the study (considered significant when p ≤ 0.05), followed by a discriminant function analysis. Analysis was performed in SPSS version 25.

| RE SULTS
Initially, 132 compounds were detected, of which most were identified. However, many were alkyl esters in such low abundance that they were not included in further analysis. Obvious contaminants were also excluded; this was the case for several identified compounds, Vitamin E (α-tocopheryl) acetate, Parsol MCX and an isomeric form of identical mass spectrum, Parsol 1789 and a number of alkyl benzoates and glycerol esters. Unidentified compounds were also excluded. The 44 compounds which therefore remained for analysis can be seen in Table 2 where the mean of the three replicates for the 44 donors was used for data analysis. A large proportion of compounds were esters and fatty acids with only three alcohols, a terpene and sterols. A full list of compounds found in the samples is given in Table S1. normal elution order is i-then a-then n-for a given C n for the type of low polarity GC column used in this study [40,41]. The free fatty alcohols present were the n-C 16 , C 18 , and C 20 homologs, no branched chain alcohols were detected. The predominant monounsaturated fatty acids in human sebum, unique to this secretion, have the double bond in the Δ6 position and it is, therefore, likely that the n-C 16:1 and C 18:1 acids found in the samples are (Z)-6-Hexadecenoic acid (sapienic acid) and (Z)-6-octadecenoic (petroselinic acid) acid [42][43][44][45]. The C 18:2 acid is thus probably (Z)-5, (Z)-8-octadecadienoic acid (sabaleic acid), also unique to human sebum, which is formed biosynthetically by a C 2 elongation of the precursor sapienic acid followed by insertion of the second double bond between C5 and C6.
Saturated esters consisted of both n-saturated C 28 -C 34 homologs, i-branched even carbon C 28 , C 30 , and C 34 homologs and a-branched odd carbon C 29 , C 31 , and C 33 homologs. In addition, branched (br-) odd carbon C 29 , C 31 , and C 33 components of undetermined branching point were also present. The presence of i-and a-branched free acids and apparent absence of branched alcohols suggests that in the esters the branching occurs in the acid moiety.
From examination of the mass spectral fragmentation patterns of the esters, it was apparent that for an ester of a given carbon number (C n ), several homologs were usually present with different acid/ alcohol combinations, but with the same overall chain length (CL).
The isomeric distribution within each overall CL was not determined in detail, although the identities of the homologs within each CL are listed in Table S1. Acid (n-, i-) and alcohol (n-) moieties in the range of

TA B L E 2 (Continued)
Unsaturated esters consisted of n-C 30:1 , C 31:1 , C 32:1 , C 33:1 , tually the only acid moieties in each ester with alcohols in the range C 14 -C 18 and C 20 , and that the branching was in the acid as found for the saturated esters. The sterol, cholesterol, and its biosynthetic precursor the terpene, squalene, were both present.
The overall distribution of compounds found within the fingerprint samples, by class and structure, was in agreement with that reported for sebaceous gland lipids [42][43][44][45]. Examples of chromatographic traces for samples from a female and two males are shown in Figure S1.  Table 2 and Figure 2.
All three alcohols from Table 2

| DISCUSS ION
The analysis showed that of the 44 compounds identified in latent fingermarks using GC-MS analysis using the method described above, only two compounds, both alcohols, had quantities that allowed differentiation between sexes with a statistical significance of less than 0.05. This result provides a valid avenue for further research into setting up a method that may be of use for establishing certain aspects of a donor profile from latent fingermarks. Profiling a donor using latent fingermarks could be applied in cases when fingerprints available at potential crime scenes are not conducive for identification purposes from their pattern. Certain fingerprints may be smudged or consist of only a partial print or overlap with other prints hence distorting the pattern and may therefore not be appropriate for identification from the ridge pattern. In such cases, the fingermark may be used to retrieve information such as the sex of donor where fingerprint visualization techniques and destructive nature of GC-MS analysis allow for its application [46].
Latent fingermarks degrade after deposition, however, it has not yet been possible to establish a rate of degradation [19]. For this F I G U R E 1 Estimated means with error bars (95% CI) for octadecanol C18 (compound 2 in Table 2). research, it was not possible to collect all the samples at the same time and avoid a certain level of degradation. Samples from different donors were collected with some hours in between. However, to limit the effect of fingermark degradation the samples were analyzed the same day as collection. Further research is needed to establish a rate of fingermark degradation in order to make it more applicable to crime scene scenarios [19].
Fingerprints collected from a crime scene would most likely have higher levels of contaminants and hence could exhibit different chemical compositions compared to the results from this F I G U R E 2 Estimated means with error bars (95% CI) for eicosanol C20 (compound 3 in Table 2).

F I G U R E 3
Box plot for male and female values of compound number 2, octadecanol (C18) (outliers and extreme outliers are included in analysis but not shown for values above 0.08).
research. Even newly washed hands will be contaminated by fatty acids from soap which are likely to contain some of the same fatty acids as found in this study [47]. However, it was anticipated that when the donors wiped the skin behind the ears to enhance analyte levels, this would to an extent mimic the residue somewhat to be found on 'dirty' fingerprints. It was initially hoped that wiping behind the ears, rather than the face as other studies have done, would eliminate contaminants such as make-up, moisturizer, and shaving foam. However, some obvious contaminants, such as Parsol MCX (2-Ethylhexyl-4-methoxycinnamate, two forms likely cis and trans positional or stereo isomers), Parsol 1789 (4-t-butyl-4 ′-methoxydibenzoylmethane), and alkyl (C 12 -C 15 ) benzoates were found in this study. Parsol MCX is an ingredient in skin care products, predominantly as a UVB filter [48] and Parsol 1789 is a UVA absorber. Alkyl (C 12 -C 15 ) benzoates are emollients and are found in many personal care products. The antioxidant Vitamin E (principally α-tocopherol) is a constituent of sebum secretions and it has been suggested that sebum may serve to deliver α-tocopherol to the skin surface where it functions as the main skin antioxidant [43]. However, Vitamin E acetate (α-tocopheryl acetate) is an ingredient in skin care products as an anti-aging product [49], con- This study only included adults with an age above 20 years, as this is when puberty is considered to have ceased [50,51] and thus there should be less fluctuation in hormone levels which have proven to be evident from fingermark residue [28,52]. Individuals receiving hormone or steroid treatment for illness, ailments, or for the purpose of gender affirmation [53][54][55] were also excluded as it was uncertain how this might influence fingermark residue.
Androgens are known to increase secretion of sebum, whereas estrogens can have the opposite effect. Uptake of circulating lipids is a significant step in the production of sebum lipids. However, some constituents such as wax esters which are unique to sebum and sebaceous cells is not converted to cholesterol as happens readily in other tissues [43]. The majority of the compounds formed in the sebaceous gland are the products of acyl chain synthesis and chain modification, fatty acids being direct products and alcohols being formed by subsequent reduction steps. It remains unclear how steroid or other treatments might affect these processes and the balance between different products. Chemical inhibitors and mutations can affect these processes and some of the steps may be influenced by transcription factors. This is usually seen as a change in chain length distribution within a compound class or a block in the formation of specific classes or a change in the distribution of different compound classes.
It was not possible to exclude females taking oral contraception, rather it was presumed that the majority were. It was also noted if the donors used asthma inhalers as some of these contain steroids.
This was only the case for three individuals. However, none of the three individuals utilized "prevent inhalers" which are used every day to prevent an asthma attack. Prevent inhalers use steroids as the main drug. The three donors all used "reliever inhalers" to relieve symptoms such as breathlessness. The main drugs for reliever inhalers are bronchodilators [56]. Therefore, for these three individuals, no difference was expected to be apparent, and none was sought.
Certain outliers in the data were noticed with particularly one female for compound 1 having a much higher value (the maximum of 0.7943, Table 2) than the female with the lowest value (the minimum of 0.0019, Table 2). However, as all three replicates for this female showed equally high values (the same were true for other outliers) it was considered that outliers were a natural part of human variation for fingermark residue components. This was confirmed by the studies of [23,57] who both found reproducible outliers which could be explained by variability in skin surface secretion due to differences in age, sex, race, disease, and activity levels. It was therefore decided not to manipulate the data by deleting certain outliers.
While some studies on the chemical analysis of latent fingermarks were not able to distinguish between males and females [34,35,37] others did find equally encouraging results [6,8,11,36,38].
The study by [6] analyzed fingermarks residue from 18 donors, nine males and nine females, using GC-MS. They found that mean levels of amino acids where higher in females but only statistically significant for asparagine, while the mean level of most fatty acids where higher in males although not at a level that was statistically significant. In our study, we found similar results for fatty acids, with mean male values higher than females ( Table 2), and equally found this to be statistically non-significant.
The use of liquid chromatography-mass spectrometry (LC-MS) [8] allowed for analysis of the much less abundant amino acid components from fingermark residue which were not detected in our study. They found that the amino acid composition between female and male donors was comparable, although for serine there was a distinct difference with much higher levels in males. The study by [11] using GS-MC analysis only mentions that the quantity of urea is sex-dependent but provides no values or number of donors in their study as this was a report on method development for fingerprint extraction using GC-MS analysis and not a study of the chemical components themselves.
Michalski et al. [38] likewise used GC-MS for the analysis of the fatty acids of fingermark residue of 22 male and 15 female donors, all within an age range of 18-21 years. They found higher levels in males of (Z)-6-octadecenoic acid while higher levels of octadecanoic acid were found for females. They also found higher levels of fatty acids from females with chain lengths intermediate between C 21 and C 22 , C 18 and C 19 , and higher levels in males of acids with chain length intermediate between C 16 and C 17 . Since they used a GC column (Rxi-5Sil MS) of similar polarity with similar elution characteristics to that used in this investigation, it is likely that these acids were the branched i-C 22 , a-C 19 , and a-C 17 homologs, respectively. However, the statistical significance between the sexes for all these observations was not reported. They analyzed the fatty acids as methyl esters after chemical derivatization. However, they did not quantify their data in the same way as used in our study, that is, based on a fixed amount of internal standard added to each sample. Instead, i-branched unsaturated esters (C 32:1 -C 34:1 ) and squalene were also greater for females. There are therefore similarities between our findings, and those of [38], particularly in finding higher levels of (Z)-6-octadecenoic acid (n-C 18:1 ) and a-C 17:0 from males and higher levels of i-even C acids from females (i-C 22 , i-C 16 ).
The potential for differentiation between sex in the distribution of i-and a-branched free fatty acids and i-branched wax esters, as suggested by our results and [38] is worthy of further investigation. Although rare in humans and apparently restricted to the sebaceous gland, it is likely that formation of the br-acids arises from parallel but separate elongation systems utilizing branched primer units (isobutyryl-CoA for the i-even C series, 3-methylbutyryl-CoA for the i-odd C series, 2-methylbutyryl-CoA for the a-odd C series), as is the case in plant cuticular waxes [58,59]. Unbranched n-fatty acids arise from elongation of acetyl-CoA (even C acids) and propionyl CoA (odd C acids), possibly also from separate but parallel elongation systems. The C 2 elongating units are primarily malonyl-CoA derived from acetyl-CoA. A branched elongating unit, methylmalonyl-CoA, can be incorporated either to give a-odd C compounds or internally branched compounds.
Acetyl-CoA used in sebaceous glands for fatty acid synthesis, particularly in wax ester formation, is formed primarily by β-oxidation of linoleic acid extracted from circulating lipids [43]. There is therefore the possibility for differential expression of products from these pathways influenced by various genetic and environmental factors, which might affect the structural and chain length distributions between and within the n-and br-classes. This is similar to the ways in which cuticular wax formation in plants is strongly influenced by genetics and interaction with environmental factors such as exposure to pollutants and mutagens, irradiation levels, temperature, and humidity [58,59].
Although our study was performed on a similar limited number of donors, our study is also in agreement with some of the other results by indicating the value of chemical analysis of fingermark residue for donor information such as sex.

| CON CLUS ION
Within the 44 compounds identified in this study, two alcohols, octadecanol C 18 and eicosanol C 20 , were found to show a difference that was statistically significant between male and female donors.
This result indicates that these two compounds could potentially be analyzed to provide information about the sex of the donor in cases when unknown latent fingermarks cannot be used for identification from ridge pattern comparison. There is also some evidence for the possibility of distinguishing sex from a latent fingermark based on the distribution of branched chain fatty acids, as free compounds or esterified in wax esters.

ACK N OWLED G M ENTS
The authors thank CAHID (Centre for Anatomy and Human Identification) at the University of Dundee for support during this project. Gratitude is also extended to Dr Stephen Hubbard from the University of Dundee for statistical advice and Professor Caroline Wilkinson from Liverpool John Moore University for guidance.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflict for the publication of this manuscript and no funding was involved for this study.