Introduction
Materials and methods
Eligibility criteria
Search criteria and critical appraisal
(“forensic microbiology” OR “microbial forensics” OR “forensic microbial analysis” OR “microbiological evidence”) AND (“human identification” OR “biological identification” OR “identity determination” OR “forensic identification”).
Risk of bias
Results and discussion
Authors, Year, [Ref] | Study Sample | Main Findings | Limitations |
---|---|---|---|
Fierer et al., 2010 [8] | -For the keyboard study, three pc and their owners, 15 other private and public pc -For the computer mice study, 9 individuals and 270 other hands sampled | -Greater similarity between personal objects and owners (significantly for the computer mice study) -No significantly influence on storage under typical indoor conditions for up to 14 days | -Sample size -Individuals shared office space/same buildings -Sample and storage conditions far from reality (for example control environmental) |
Schmedes et al., Sept. 2017 [9] | -14 skin body sites from 12 individuals sampled at 3 time points over a > 2.5-year period | -The manubrium site and the hypothenar palm yield highly accurate rates of classification (97% and 96% respectively). Nucleotide diversity of stable markers yielded accuracies as high as 100% from some body parts and contributed significantly greater to classification accuracies (p < 0.01). | -Sample size -Low intraindividual samples -Uncertain applicability to real forensic applications |
Schmedes et al., Oct. 2017 [10] | -8 individuals, swabs from 3 body sites (foot, hand and manubrium): 3 replicate samples collected from each body site ◊ N = 72 | -Classification accuracies were highest for Hand (average 97.9% SD 2.1) than for Manubrium (average 86.3% SD 6.9) using universal markers. Their accuracy calculated using enriched markers were comparable (p = 1). -Classification accuracies for foot were significantly higher (p < 0.00001) using enriched markers (up to 92%) than universal markers (up to 23%). Body site origin was predicted with up to 85% accuracy. | -Laboratory bacterial contamination -Microbiome shared by individuals (for example cohabiting couples and family members). -Stability of skin microbiomes collected over time intervals -Further body sites and markers of bacterial genus |
Watanabe et al., 2018 [12] | -11 individuals, 66 samples for 2 years (33 for each year) -A public dataset (837 publicly available skin microbiome samples) of skin microbiome samples from 89 individuals | -Personal identification accuracy of 95% (63/66); using three reference samples from the first year and three query samples from the second year, they found accuracy of 85% (28/33). -Using a public dataset (89 individuals), they calculated a personal identification accuracy of 78% (663/837). | -The need of extensive dataset of sampled skin microbiome and of more studies on temporal dynamics of the microbiome |
Park et al., 2017 [11] | -15 individuals, 696 bacterial strains isolated from palm hand | -They found personal variation of some minor species and hypothesized the major species could apply as molecular biological markers | -Sample size -Features of the individuals, without explaining how these factors could influenced the results (for example, taking of antibiotics). |
Meadow et al., 2014 [15] | -51 samples from 17 participants (3 samples for each individual) | -They mainly found that an individual’s finger shared on average 5% more OTUs with his or her own phone than with everyone else’s phones (p < 0.001). | -Environmental conditions -Sample size -Design of study (teaching exercise) -Type of phone (only smartphones without keypads) -Hand-washing methods variable not considered |
Neckovic et al., 2019 [13] | -6 individuals placed into 3 pairs for a total of 65 samples | -The Jaccard and unweighted Unifrac distances between samples, revealed that the clustering of participant pairs was distinct. | -The need of negative controls -Bacterial contaminations -Sample size |
Lax et al., 2015 [14] | -2 individuals (total samples 315) -Geographically study: 89 individuals (unknown total number of samples) | -They assessed that phone-associated microbial communities were observed to be both less stable (higher median distance) and more variable in their rate of change over time (broader distribution) than shoe-associated communities. | -Rapid turnover of the surface-associated microbial community -Sample size -Short sampling time (two days) -Few substrates analyzed |
Costello et al., 2009 [16] | -For the first study: 815 samples (7–9 adults on four occasions) -For the second study: n = 16 | -Within habitats, interpersonal variability was high, while individuals exhibited minimal temporal variability. -Environmental characteristics play a strong role In shaping skin bacterial communities | -The need of a longer observation period to assess the influence of environmental factors and historical exposures |
Phan et al., 2020 [17] | -45 individuals (left and right hand were obtained from each subjects: total of 90 samples | -They found Alloiococcus species could be a potential biomarker for sex (64% accuracy rate) and ethnicity (56% accuracy rate) | -Large standard deviation in samples Lack of clear distinction between groups -Low robustness of predictive models (apart from sex prediction) -Low prediction accuracy rates -Small sample size uneven breakdown of groups within each examined factor considering other substrates than playcards -Bias of individuals: same group (university) and not considering if they were co-habitants or if they had pet -Single site point: subsequent sampling |
Richardson et al., 2019 [18] | -37 individuals | -They predicted the sex of the individual (error ratio of about 2.5, and accuracy of around 80%) | -Population of students living in the same dormitory -The bias of presence of roommates |
Fierer et al., 2008 [19] | -51 individuals, 102 samples from 27 M and 25 W | -Hands from the same individual shared only 17% of their phylotypes, with different individuals sharing only 13%. This intraindividual differentiation was not significantly affected by handedness, sex, or hand hygiene (P < 0.05 in all cases). Men and women harbor significantly different bacterial communities on their hand surfaces (P < 0.001). | -Sample limited to students -The lack of detailed information on the skin characteristics of the sampled individuals. |
Bell et al., 2018 [20] | -10 cardiac tissues | -sex-dependent changes in the thanatomicrobiome composition were statistically significant (P < 0.005). | -Small sample size -Analysis of the variability of the bacterial community based on the time elapsed since death -Other substrates into consideration |
Tridico et al., 2014 [21] | -Forty-two pools of DNA extracts from 7 individuals (4 F and 3 M) | -Lactobacillus spp. was found in the female pubic hair samples and not in the male samples (excepting in the cohabiting male). Instead, similar microbial taxa were observed in the cohabiting couple, suggesting interindividual transfer. In contrast to the pubic hairs, scalp hair microbiota showed no correlation with the sex of the donor. Moreover, pubic hair microbiomes appeared to be less influenced by environmental bacteria than scalp hairs. | -Sample size -Scientific validation including: replication, persistence of bacteria during contact and stability during storage |
Pechal et al., 2018 [22] | -83 individuals | -Decreased phylogenetic diversity was observed as a significant predictor (P = 0.038) of heart disease | -Age bias of individuals |
Nagasawa et al., 2013 [23] | -10 individuals | -Phylogenetic tree of H. pylori showed 3 major clusters. All of the Japanese (n = 10), South Korean (n = 1), and Chinese (n = 2) cadavers examined in the present study were classified as type I, the single Thai cadaver was classified as type III, and the single Afghan and Filipino-Western cadavers were classified as type II. Different classification in this study could be due to external factors (i.e. latent origini) | -Results influenced by the latent origin -More geographical origins and knowing the background details of the analyzed sample, mostly unknown in this article |
Escobar et al., 2014 [24] | -Total individuals 126, 30 Colombian adults | -The UniFrac analysis indicated that the gut microbiota of Colombians was significantly different from that of Americans, Europeans and Asians (P = 0.001). Moreover, they found that the relative abundance of Firmicutes decreased with latitude (r = − 0.27, P = 0.002) and that of Bacteroidetes increased with latitude (r = 0.28, P = 0.001) | -Sample size -No statistical power (no previous data on Cloumbians, high variability) -Influencing environmental factors (i.e. diet) |
Brinkac et al., 2018 [25] | -Total individual of 21, for a total of 42 and 32 hair samples from scalp and pubis respectively. | -Scalp hair showed greater potential to predict geolocation than pubic hair | -Influencing factors: hair length, sebum content, lifestyle factors -Sample size -The need of longitudinal studies |
Ghemrawi et al., 2021 [26] | -10 genital samples (10 individuals) | -They found that the penile microbiome species composition was different from vaginal | -Sample size -The need of longitudinal studies -The sample not included couples neither information regarding recent intercourse -Preliminary study |
Williams et al., 2019 [27] | -43 individuals, 155 total sample collections | -A correlation between the proportion of couple co-clustering and the frequency of sex intercourse Increased frequency of sexual activity didn’t however guarantee increased microbiome similarity | -Larger sample sets -The need of controlled studies |
Dixon et al., 2023 [28] | -6 male-female sexual partners pair, 20 swabs for couple | -Both the male and female genital microbiomes might be susceptible to alteration by the opposite sex | -More specific details (specific intimate behaviors, menstruation, health status, time of sampling) -Larger study group |
Kennedy et al., 2012 [29] | -16 individuals | -The 16 S rRNA model revealed a sensitivity of 100%, with a 25% false positive rate. The ITS model found a 65% chance of obtaining a false positive. Finally, the rpoB model matched all bite marks to the corresponding teeth | -Sample size -Self-produced bite marks -Influencing factors (for example dental diseases) -Difficult adaptation to reality |
Personal microbiome and transfer to the surrounding environment
Microbiome as indicator of biological profiling features
Geolocation
Determination of sexual contact
Limitations
-
the instability of the microbiome to intrinsic and extrinsic factors, for example the use of antibiotics or the presence within the subject of a certain disease or hormonal factors that modify the microbiome;
-
the difficulties in maintaining the ideal conditions in carrying out the sampling, transport, and treatment of the microbial community: in fact, different microbial populations may require different protocols. Furthermore, according to what is accepted by the scientific community, a valid protocol should have been tested in field conditions, whether it has been subjected to peer review, whether the rate of error is known, standardization and whether it has been generally accepted. This scientific methodology is the way to ensure the reproducibility and comparability of research results to be applied in concrete cases of judicial investigation (with the same safeguards in terms of privacy and confidentiality as any other human tissue samples or identifying sources of information);
-
the presence of contaminants of human, environmental and other living beings, such as animals and insects;
-
the sample size of the studies presented in this review, which appears to be still too small;
-
the standardized definition of changes in the microbial community in the postmortem period. The postmortem human microbiome, such as Javan et al. reported, includes two components: the thanatomicrobiome, consisting of microbes that inhabit internal organs and body fluids after death, and epinecrotic microbial communities, represented by microbes found on the surface of decaying remains [4]. In fact, the thanatomicrobiome is conditioned by many endogenous and exogenous factors, including climatic conditions and the presence of animals, as well as by postmortem translocation and agonal diffusion phenomena;
-
almost all of the studies presented (with the exception of the study by Kodama et al. [30]) were constructed according to a rigid design to control interfering variables (e.g., hand washing in subjects, voluntary recruitment of subjects). Nonetheless, such a study design could be difficult to adapt to real forensic applications;
-
many studies should be conducted to identify a common matrix (i.e. a sampling site) less influenced by external and internal factors. For example, hands represent a useful site because they are more involved in contacts, but also more susceptible to confounding factors. On the contrary, for example the forehead is less susceptible to external contacts but influenced by individual factors (for example sebum production);
-
robust information on the stability of the microbiome over time is lacking. Some studies included in this review explore these differences by including some time set points. However, information on baseline time and long term is often lacking.