New approaches in estimating the post-mortem interval (PMI) – The human microbiome as an indicator of the time since death

(1)

New approaches in estimating

the post-mortem interval (PMI) –

The human microbiome as an

indicator of the time since death

Magdalena Birkl

11406739

Master Forensic Science

Supervisor: Prof. Dr. Roelof-Jan Oostra

Co-assessor: Dr. Annemieke van Dam

Date: 24.2.2018

(2)

2

List of abbreviations

ADD Accumulated degree days

C.perfringens Clostridium perfringens CDH Cumulative degree hours

DNA Deoxyribonucleic acid

MALDI-TOF Matrix-assisted laser desorption/ionisation time of flight mass spectrometry

MLN Mesenteric lymph node

NGS Next generation sequencing

PMI Post-mortem interval

qPCR Quantitative polymerase chain reaction

rRNA Ribosomal ribonucleic acid

RT-PCR Reverse transcription polymerase chain reaction

S.aureus Staphylococcus aureus

(4)

4

Abstract

The determination of the Post-Mortem Interval (PMI) is a crucial step in a death investigation. Several currently used methods however are prone to errors and result in large uncertainties in time since death. As microbes are ubiquitous in nature and as the human microbiome is well studied in the living human it is considered to play a major role in the human decomposition process. A few studies have been conducted up to now to investigate the microbial migration throughout the decomposition process with means of animal models, such as pig and mice, and human cadavers. The investigation of external and internal sampling sites during the decomposition process revealed a distinctive shift from aerobic to anaerobic bacterial communities. However, also several factors which could influence the accuracy of the proposed method were identified, for example inter-individual and sex-specific variations as well as seasonal variations. Nonetheless, 4 study groups were able to establish a prediction model for the estimation of the post-mortem interval (PMI). Due to the small sample size and many limiting factors, additional research is recommended.

(5)

1. Introduction and background

1.1. Research question

The main focus of this literature thesis lies on discussing the current status in the field of post-mortem interval (PMI) estimation using the human microbiome. Questions that arise in this context are for example to which extent the human microbiome is involved in the decomposition process itself. In addition, it has to be evaluated if a method which uses the microbiome adds value to the currently used techniques and is able to estimate the PMI more accurate.

1.2. Background

The determination of the PMI is a crucial step in every death investigation since it helps identifying suspects and the confirmation and rejection of alibis (1). Therefore, a reliable and accurate method is necessary to assess the PMI and to use the respective results as physical evidence before court. Currently several standard methods are t used for the PMI estimation having all specific limitations These limitations are the trigger for studying a new approach for the PMI estimation. The time since death can be evaluated by various methods which focus on different aspects of the decomposition process. One well known method is based on findings in the field of entomology (2). With help of insects and the deposition of their eggs, the PMI can be established. Each insect species is differently attracted by the cadaver and will place their eggs at specific points in time depending on the decomposition status. The development of the insect larvae is extensively studied and can be used as a time indicator. However, placement of insect eggs is dependent on the accessibility of the cadaver and on insect activity, being dependent on temperature, weather conditions, seasonal variations and geographic location (3). This leads to possible inaccuracies if the access to the body is not possible or the egg deposition is delayed due to reduced insect activity, for example in the winter season (4).

Another often used method is the mortis triad which consists of algor mortis, livor mortis and rigor mortis (2). This method focuses on the settling of the blood in the lower parts of the body (livor mortis), the stiffening of the body post-mortem (rigor mortis) and the decrease of the body temperature after death (algor mortis). All 3 tests are prone to inter-individual variation and are in the case of livor and rigor mortis only of a temporal duration. For the estimation of the PMI with the means of the body cooling (algor mortis), Henßge developed a nomogram which is considered as the ‘gold standard’. However, also in this case different factors limit the accuracy. The body needs to be physically recovered from the place of death and the person could have an altered body temperature prior to death (e.g. fever) (2). It is also possible to measure the change in the concentration of potassium in the vitreous humour which also has a considerable error rate (4). This leads to the conclusion that a new approach for the estimation of the PMI is necessary, which could support the other methods. The new approach needs to be repeatable and predictable and has preferably a lower error rate. One option which is considered for this is to use the human microbiome.

Generally, the human microbiome is considered to be a major player in the decomposition process of soft tissue (4-5). Microbes are ubiquitous in nature (6) and the processes of human decomposition mediated by the microbial communities instantly begin after the death of the host (7). As the human immune system is no longer functional after death, it helps with the proliferation of the microbes. The nutrient-rich environment of the cadaver additionally supports the proliferation process (8). Furthermore, it is assumed that the microbial communities of the body changes after the death of the host which could be used as an indicator for the PMI (4).

Due to the before mentioned limitations of the currently used methods, it is necessary to conduct research to find new approaches for the PMI estimation. The human microbiome is a possible source

(6)

6 of information for the PMI. It is well studied in the living human, but little is known about the post-mortem processes which are related to the microbiome (9). If the human microbiome would change in a repeatable and predictable manner, it could be used as an indicator for the PMI (10).

(7)

2. Overview studied literature

The studied literature contains 19 research papers (see Appendix), which can be divided and analysed in several categories. The papers were published between 2012 and 2017 and discuss the bacterial microbiome on and in decomposing human bodies and animal models. In the scope of this thesis, the literature is divided in the categories human and non-human cadaver studies.

2.1. Human cadaver studies

In this category, 12 different papers were analysed and compared according to several criteria. Each paper studied the decomposition process under natural conditions, either in an outdoor research facility (3-5,10-11,14-16) or using cadavers who died of natural or unnatural causes outside of the hospital (8-9,12-13).

2.1.1. Sample size

The sample sizes varied from 2 (5,11) to 45 (12) human cadavers. In the 2 studies conducted by Hyde et al. (5,11) the cadavers were frozen for a significant time before the outdoor placement and the initial PMI was unknown to the researchers. In one of these studies (11) the medical history of the investigated cadavers was not known. The remaining 10 studies used bodies with a known PMI which ranged from 4 hours (12) to 1692 days (14) representing the complete decomposition process from the earliest stage until the stage were only dried remains are left. The sample size of these studies ranged from 3 to 45 human cadavers.

2.1.2. Sampling points

The amount of sampling points is varying and depending on the fact if the study is done in an outdoor research facility or not. During a study at an outdoor research facility it is possible to place human cadavers and follow the decomposition process under chosen conditions. In this case it is possible to sample the bodies several times at specific points in time. Most studies have chosen a 1 to 2 days sampling interval (4,10-11,15-16) or an interval of 2 to 3 days (3). An exception to this are the studies of Damann et al. (14) and Hyde et al. (5) which sampled only at 1 respectively 2 points in time and not continuously over a period of days or weeks. Studies which were performed outside an outdoor research facility only had 1 sampling point. The sampling interval and date also influences the sampling regions which are discussed in the next chapter.

If control samples are considered in the study, e.g. in the study of Hauther et al. (4), these are usually only sampled once. When more than 1 control sample is present, it is possible to choose several points in time.

2.1.3. Sampling region

The different research groups have chosen several different internal and external sampling regions and give diverse motivations for each choice. It is notable that the 8 studies which analysed the whole natural decomposition process outdoor (3-5,10-11,14-16) usually choose easy accessible regions like the oral cavity (5,11,16), the skin surface (11,15), the rectum (5,11), the nasal and ear canals (3) and rib bones from (partially) skeletonised remains (14). An exception are the studies of Hauther et al. (4) and DeBruyn and Hauther (10) which analysed the gut microbiome via accessing the body through an incision in the abdomen.

Johnson et al. (3) also motivated their choice for an externally accessible sampling area with the advantage that the body is not disrupted by the sampling at the crime scene. On the other hand, the human gut microbiome is considered to be very consistent in nature despite individual variations due to diet and lifestyle (4). The time-dependent changes of the gut microbiome already have been studied in living humans and Hauther et al. (4) tried to quantify changes in the gut microbiome after the death of the human host.

(8)

8 The remaining 4 studies used bodies from criminal cases or who died outside a hospital. Therefore, it is possible to analyse internal body sites which can be accessed during an autopsy of the bodies. The sampled regions encompass the liver and spleen (8-9,12-13), brain, heart and blood (8-9) as well as the mesenteric lymph node (MLN), pericardial fluids and portal and peripheral vein blood (13). The choice of internal sampling sites is also reasonable, since internal organs are considered to be sterile during life, unless there is a bacterial infection which could be life-threatening (13). The tissue of internal organs is also less affected by the environment compared to external sites like the skin or oral mucosa (9).

An overview of the sampling regions can be found in table 1.

Table 1 Overview of the human cadaver studies and the associated sampling regions for each study

Study Year Sampling region

Hyde et al. (5) 2013 Pre-bloat samples: mouth & rectum; internal organs

Tuomisto et al. (13) 2013 Liver, MLN, pericardial fluid, portal & peripheral vein

Can et al. (9) 2014 Spleen, liver, brain, heart, blood

Damann et al. (14) 2015 Single lower rib (skeletonised bodies)

Hauther et al. (4) 2015 Gut microbiome

Hyde et al. (11) 2015 Mouth, cheek, bicep region, torso, faecal

Javan et al. (8) 2016 Brain, heart, liver, spleen, buccal cavity and/or blood

Johnson et al. (3) 2016 Skin microbiota of nasal and ear canals

Metcalf et al. (15) 2016 Winter: 3 skin (left and right hip, knee), 4 soil; spring: 8 skin sites, 6 soil sites

Adserias-Garriga et al. (16) 2017 Oral cavity (palate, tongue, inner cheek mucosa, tooth surfaces)

DeBruyn and Hauther (10) 2017 Gut microbiome

Javan et al. (12) 2017 Liver, spleen tissue

2.1.4. Sampling technique

The applied sampling techniques can be divided into destructive and non-destructive techniques. All non-invasive external sampling sites were swabbed (3,5,11,15-16), as were the organs extracted during an autopsy in the study of Javan et al. in 2016 (8).

In the first study of Hyde et al. (5), they also used a spatula for “scraping” the surface of mouth and rectum and compared the results to those obtained by swabbing. Results showed variation in the relative abundance of bacteria in one body between these 2 applied sampling techniques (swabbing and scraping) which indicates an influence of the used sampling method on the obtained bacteria and their relative abundance.

The 2 studies which sampled the gut microbiome (4,10) also swabbed the intestines, but accessed it via a small incision in the abdomen and sealed it with duct tape afterwards. For re-sampling, a small incision was made in the tape. The tape was left to reduce insect activity until natural decomposition processes prevented the tape from sticking (4).

Hauther et al. (4) used 6 bodies as control samples for their chosen sampling technique to check whether repeated sampling has an influence on the bacterial population in the gut. Results showed a possible effect of repeated sampling on the relative abundance of the analysed bacterium

Bifidobacterium. For the other 2 analysed bacteria Bacteroides and Lactobacillus the control samples

(9)

and Hauther (10) concluded from the comparison of the control and test samples that the sampling method does not have an influence on the dominant bacterial population.

One study did not disclose which exact sampling technique they used. It was described as sterile technique which can be used for culturing bacteria and performing a quantitative Polymerase Chain Reaction (qPCR) (13).

Another approach, which is destructive in nature, is the dissection of organs for further processing and desoxyribonucleic acid (DNA) extraction (12). If the body is already in a partially skeletonised or fully skeletonised stage, it is possible to collect bones for further processing and DNA extraction as done by Damann et al. (14).

It is important to mention the possible influence of the sampling technique on the results as discussed by the studies of Hauther et al. (4) and Hyde et al. (5). This topic is further detailed in the next chapter.

2.1.5. Analysis method

The first step of the analysis is the DNA extraction process. Can et al. (9) provided a comparison of 2 different extraction methods, namely lysis of pieces of organ tissue (method 1) and swabbing of the same organ (method 2). After the subsequent PCR amplification, the presence of products was compared and indicated that method 1 yielded more PCR products than method 2. The reason for this could be the extra step to remove debris during the analysis. The research group concluded that method 1 is preferred over method 2 (9). This comparison also implicates, that the extraction method could have an influence on the results and should be evaluated before the actual analysis. The analysis methods can be divided in a culture-based and culture-independent methods. Today it is most common to use culture-independent methods and amplify the 16S ribosomal ribonucleic acid (rRNA) to identify bacterial phyla. This was done by 10 studies (3,5,8-12,14-16).

The study of Tuomisto et al. (13) is the only study which used the culture-based approach for the analysis with a subsequent qPCR. The study of Hauther et al. (4) also used qPCR to quantify 3 common gut genera. In this case it is not obligatory to analyse the whole present bacterial community, but focus on the chosen genera. For each of the 3 genera, a specifically designed assay was used.

For the subsequent sequencing step several hypervariable regions were analysed. Previous studies have shown that the hypervariable region V4 is the best region for phylogenetic studies (12). For the analysis of potentially forensically relevant bacteria, Javan et al. (12) determined that the hypervariable region V4 has a higher sensitivity than the compared V3 region. Every study which used 16S rRNA for the analysis included the V4 region in their analysis.

The results of the analysis are discussed in the following part.

2.1.6. Results and conclusions

The results are mainly based on the presence or absence of certain bacteria and their change in relative abundance.

The study of Hyde et al. (5) recognised a shift from aerobic to anaerobic bacteria during the process of decomposition. Clostridium was present in a large number at the end of the bloat stage in most of the sampled locations. But they also observed a big difference in detected bacteria between the 2 studied bodies and also between body sites in the same body (5).

The second study of Hyde et al. (11) observed changes in the structure of the bacterial community for all body sites sampled during the process of decomposition. The relative abundance of bacteria in both cadavers showed similar changes over time. However, at the generic level, some variation

(10)

10 exists between the 2 cadavers which could be the effect of the freezing of the bodies for a different amount of time (11). Across all body sites, the bacteria Ignatzscheria and Actinetobacter dominated one body, as did Clostridium and Actinetobacter in the other. These bacteria are associated with myasis by fly larvae, soil and animal and human feces and could be key contributors to the general decomposition process (11). During the process of decomposition, the positioning of the body may change when, for example, the head turns at a specific point in time. Hyde et al. observed a difference in diversity and relative abundance of taxa which are present on the left and right side of the body and which could be explained by these events (11).

Javan et al. (12) observed differences in species richness when comparing sex and cause of death. The research group proposed a new scientific concept which is called the “Postmortem Clostridium Effect” and refers to the omnipresence of the species throughout the whole decomposition process (12). The bacterium is able to digest vertebrate collagen fibres which aids in the process of transmigration to other tissues. It is also an anaerobic bacterium which is naturally found in the colon and profits from the cessation of the human heart and the resulting hypoxia (4). This shift from an aerobic to an anaerobic environment and the profitable change for the Clostridium has previously been reported and is also supported by this study (12).

The perspective of the study of Tuomisto et al. (13) is more medical than forensic. They assessed the sterility of several internal organs and body fluids in humans with a known PMI. The findings show that the liver can have a high sterility of up to 5 days (13). It is also observed that the relative abundance of intestinal bacteria in internal organs increased with the PMI. On the other hand, the relative abundance of intestinal bacteria in blood decreased, whereas the bacteria Staphylococcus and Streptococcus increased in their relative abundance (13). However, it must be mentioned, that the study used culture-based techniques which can only detect cultivable bacteria. Nevertheless, it is worthwhile to perform sterility assessments in further studies when repeated with a culture-independent method, supporting the estimation of early PMIs (13).

The study of Can et al. (9) supports the theory of a shift from aerobic to anaerobic bacteria. A change from predominant facultative anaerobic bacteria, like Lactobacillus, in cadavers with shorter PMIs to obligate anaerobic bacteria, like Clostridium, in cadavers with a longer PMI was observed during the studied time period (9). These results are supported by several other studies, like the study of Hauther et al. (4), Hyde et al. (5) and Javan et al. (12). Another observation was that the post-mortem microbiome varies by PMI rather than by organs (9) which is partly contradictory to the statement in several other studies, that they observed a location-dependent microbial activity (5-6,18-19).

Damann et al. (14) focused on the bacterial communities which can be found on bones of partially and fully skeletonised bones which represents the very last stage of the decomposition process. Over time, the bacterial community of the skeletal remains became more similar to the microbiome of the surrounding soil which was also partly observed by Hyde et al. (11). Proteobacteria was the most dominant phyla across all samples. The group of Alphaproteobacteria increased over time while the group of Gammaproteobacteria decreased. In the later stage of decomposition, the soil associated groups of Actinobacteria and Acidobacteria were more prevalent in the samples (14). The only exception to this pattern was one cadaver which showed adipocere formation. This effect was not observed in one of the other studied papers and needs to be mentioned as a possible influencing factor on the decomposition process (14).

Hauther et al. (4) were able to describe the observed changes in an exponential decay model which could be used to predict the PMI for a time interval of up to 600 cumulative degree hours (CDH), which is comparable to 20 days in the studied region in the USA. During the process of the study, 3 bacteria genera were analysed. Two of these genera, namely Bacteroides and Lactobacillus, showed a decrease in relative abundance over time (4).

(11)

In the study of DeBruyn and Hauther (10) the possible effect of diseases and other extrinsic factors could be observed in the way that one body had a feeding tube right before death and showed a very different starting post-mortem microbial community (10). Unfortunately, the decomposition process of this cadaver was not further followed in the study. The remaining 3 cadavers showed a diverse and similar bacterial community which clearly changed over time to more different bacterial communities among the 3 individuals (10). The late communities include Clostridium and fly-associated Gammaproteobacteria like Ignatzchineria as previously observed by Hyde et al. (11). The bacterium Clostridium also showed the greatest ability to predict the PMI for the studied samples (10).

The results of Javan et al. (8) showed that bacterial genera were similar in different organs within the same sex, but showed a sex-specific difference. An exception to this trend was the buccal cavity (8). This was also observed in the later study of Javan et al. from 2017 (12). Genera such as

Clostridium and Prevotella possessed various different species which can potentially predict different

periods of decomposition. Firmicutes, which includes Clostridium, occurred in all comparisons which can lead to the conclusion that it could be a stable biomarker across the post-mortem microbiome (8). This is also proposed as the “Postmortem Clostridium Effect” by Javan et al. (12).

The study of Metcalf et al. (15) showed a significant change of the microbial communities during decomposition. An alignment of the cadaver microbial communities and the surrounding soil communities was observed as already proposed by Damann et al. (14). With the help of a random forest model, the research group was able to predict the PMI of the cadavers placed in the spring season with the use of the data from 2 cadavers placed in the winter season. This shows that it could be possible to use the human microbiome for the prediction of the PMI (15).

Adserias-Garriga et al. (16) observed a similar successional change, even if the 3 cadavers had different oral starting situations. It was shown that different bacterial communities appear during the different stage of decomposition (16). In the beginning of the decomposition process, the endemic oral communities of the phyla Firmicutes and Actinobacteria were most prevalent in the samples and decreased until the beginning of the bloat stage. During the bloat stage, the phylum Tenericutes, which is part of the gastrointestinal tract, appears as well as Igantzschineria, which was previously observed by Hyde et al. (11) and DeBruyn and Hauther (10) (16). After that, Firmicutes, mainly Clostridiales and Bacillaceae, increased in abundance. This can be explained with the ability of these bacteria to form spores and colonise the new environment in a fast manner (16).

Jackson et al. (3) used a machine learning approach and developed different models to predict the PMI of microbial samples. With a regression model and the data from all their samples, they were able to predict the PMI of an unknown sample with an average mean error of 55 accumulated degree days (ADD) (3). However, it has to be mentioned that only a combination of the 2 analysed body sites lead to these results. When considering only 1 sampled body site, the error rate increased (3).

In the next part the study results of non-human cadaver summarized.

2.2. Non-human cadaver studies

Another approach of investigating the microbiome during the decomposition process is the use of animal models. Advantages of animal models are the possibility to create genetically modified organisms with similar pre-conditions and the usage of destructive sampling techniques. In addition, it is possible to use a much larger sample group than in the human cadaver studies which leads to a possibly higher reproducibility of the results.

(12)

12

2.2.1. Animal model and sample size

The non-human cadaver studies incorporate 8 different studies with varying sample size and animal models. The studies used either mice (1,7,15,17) or swine (6,18-19) as model organisms. The research team of Handke et al. (20) used 3 pieces of pork loin as decomposition model which had a significant influence on the decomposition process itself. The sample size varied depending on the chosen model organisms. The studies of Pechal et al. (18) and Chun et al. (19) with 3 swine carcasses used a very small sample size. The biggest study using swine as model was the study of Pechal et al. (5) which used 40 swine in varying settings. The remaining studies used mice as model organisms with a sample size varying from 40 (1) to 120 (15) mice. The study of Heimesaat et al. (17) missed to state a sample size.

The mouse models were based on specific mice strains. Heimesaat et al. (17) chose a mice strain which is nearly identical in its genetic background and is therefore considered highly reproducible.

2.2.2. Test condition

In 4 of the studies (1,7,15,17) the cadavers were placed in a controlled environment indoor using mice as model. Both studies of the research group of Metcalf et al. (1,15) investigated the mice without further treatment. Heimesaat et al. (17) also chose controlled conditions, but additionally modified one group of mice by eradicating the murine microbiome introducing a complex human microbiome to the mice to mimic human-like conditions in the mice. The team of Burcham et al. (7) divided the mice into 3 different groups, namely the control group without treatment, infected mice without post-mortem surface sterilisation and infected mice with post-mortem surface sterilisation. The group of infected mice got a mixture of Staphylococcus aureus (S.aureus) and Clostridium

perfringens (C.perfringens) administered. S.aureus was labelled with a fluorescent protein, which

was later used for imaging purposes (7).

The 4 remaining studies (6,18-19,20) placed the cadavers or pieces of pork at an outdoor research facility. The research groups of Pechal et al. (6,18) and Chun et al. (19) choose natural conditions to follow the decomposition process. Handke et al. (20) placed the pieces of pork loin in sterile plastic boxes at the outdoor decomposition facility. Pechal et al. (6) examined 2 different settings in their study. The first setting studied the seasonal variation of decomposition during the 4 seasons of the year. The second setting considered the annual variation of decomposition and the effect of necrophagous insects on the process. For this purpose, 3 of the 6 studied carcasses were enclosed in cages to prevent the colonisation with insects during the process of decomposition. For the remaining 3 carcasses, insects had immediate access to the cadavers. The annual variation was investigated by sampling cadavers in 2 consecutive years (6).

It should be noted that the test conditions in the animal studies are much more diverse than in the human cadaver studies, since there are more possibilities in studying the process of composition using e.g. genetically modified animals.

2.2.3. Sampling points

The total sampling time of the 8 compared studies varied between 3 days (17) and up to 5 months (6). The following table 2 gives an overview of the different sampling points in each study.

(13)

Table 2 Overview of the different sampling points within the different animal cadaver studies

Study Year Sampling point

Heimesaat et al. (17) 2012 No information provided

Metcalf et al. (1) 2013 8 time points (5 mice per time point) over 48 days

Pechal et al. (6) 2013 Seasonal variation: every 3-4 days until dry stage

Annual variation: 4 times in 5 days (2010), daily for 5 days (2011)

Pechal et al. (18) 2014 4 sampling points (0, 1, 3, 5 days) over 5 days

Chun et al. (19) 2015 4h, 32.5h, 57h, 73h, 96h, 120h, 144.5h over 6 days

Burcham et al. (7) 2016 1h, 3h, 5h, 24h, 5d, 7d, 14d, 30d, 60d

Metcalf et al. (15) 2016 Every 3 days over the first 2 weeks, less frequent over 71 days in total (5 mice per soil type (3) per time point)

Handke et al. (20) 2017 Regular interval for 60 days h: hours; d: day

2.2.4. Sampling region

Like the human cadaver studies, the sampling regions varied between internal and external regions of the cadaver. Most of the studies sampled the skin of the animal cadavers (1,6-7,15,18-19), while 2 studies primarily focused on internal sampling sites like the intestines, spleen, liver and kidneys (7,17) as well as the MLN, blood (17), heart and stomach (7). An overview of the sampling regions can be seen in table 3.

Table 3 Overview of the different sampling regions with the associated studies

Study Year Sampling region

Heimesaat et al. (17) 2012 Intestinal lumen; extra-intestinal organs (MLN, spleen, liver, _{kidney, cardiac blood)}

Metcalf et al. (1) 2013 Abdominal, skin (body, head), soil

Pechal et al. (6) 2013 Buccal cavity, skin, interior anal cavity

Pechal et al. (18) 2014 Buccal cavity, skin (combining 3 areas)

Chun et al. (19) 2015 Head, limb, larval mass

Burcham et al. (7) 2016 Skin, liver, lungs, spleen, heart, stomach, 1 kidney, composition _{of intestines}

Metcalf et al. (15) 2016 Skin, abdominal cavity, gravesoil

Handke et al. (20) 2017 Whole piece of pork

2.2.5. Sampling technique

The sampling techniques of the animal cadaver studies are not as diverse as in the human cadaver studies. The used techniques are swabbing and the destructive sampling of the animals at specific points in time. This gives the opportunity to reach internal sampling sites. Other than in the human cadaver studies, it is therefore not possible to investigate a possible influence of the sampling technique on the results.

Only the study of Heimesaat et al. (17) disinfected the skin of the mice before the necropsy to reduce the number of bacteria on the skin surface and a potential source of contamination.

(14)

14

2.2.6. Analysis method

Compared to human cadaver studies, the analysis methods in the animal studies are more diverse. Three studies (7,17,19) used culture-based methods to cultivate the sampled bacteria. Heimesaat et

al. (17) further used a quantitative reverse transcription polymerase chain reaction (RT-PCR) of the

16S rRNA for 9 main bacterial groups to identify the bacterial species. The group of Chun et al. (19) used matrix-assisted laser desorption/ionisation time of flight mass spectrometry (MALDI-TOF) to identify the bacteria. If MALDI-TOF was unable to identify the species, 16S rRNA amplification was performed additionally (19). The motivation to use MALDI-TOF instead of 16S rRNA is due to the fact that it is cheaper and has a shorter turnaround time from sample collection over treatment to identification of the species. On the other hand, it has the same limitations as the 16S rRNA method as it is also limited to a reference database (19). Another study which applied the culture-based methods was the group of Burcham et al. (7). The selection of the culture-based approach in this study is reasonable since they only investigated the 2 bacteria S. aureus and C. perfringens, which are both cultivable bacteria. To be able to identify bacteria which were introduced to the organism, these bacteria were fluorescently labelled and could be observed under the microscope (7). In addition to the culture-based approach the research group performed a whole-body fluorescence imaging to be able to follow the migration of the bacteria throughout the body over time (7). The remaining studies used the culture-independent approach of 16S rRNA amplification (1,6,12,18,20). Metcalf et al. (1) further specified the amplified region with V4, while Pechal et al. (6,18) further specified the amplified region with V1-3. The research group of Handke et al. (20) wanted to test the ability of using terminal restriction fragment length polymorphism (T-RFLP) as an alternative for next generation sequencing (NGS) to analyse 16S rRNA. The advantages of T-RFLP are that it uses the same equipment as forensic DNA-profiling and is cheaper than NGS (20). The results of these analysis methods are discussed in the next part.

2.2.7. Results and conclusions

The studied papers are summarised in the following section.

Heimesaat et al. (17) divided the sample in 2 different groups. The first group represents mice which harbour the conventional murine microbiome. Generally, a time-dependent increase of the total bacterial load could be observed. However, the group also observed a bi-phasic detection pattern of the obligate anaerobic Bacteroides, Prevotella and 2 Clostridium species (17). A possible bi-phasic detection pattern could lead to difficulties in the later interpretation of the microbial communities and the estimation of a PMI. This will further be discussed in a later section. Additionally, the culture-positivity of the samples of the sterile internal organs was assessed. Already within 5 minutes, over 50% of the MLNs showed a positive culture which indicates a rather quick spreading of the microbial communities after death (17). This is contradictory to the results of Tuomisto et al. who observed a sterility of organs for up to 5 days post-mortem (13). An explanation for this could be the different sample of mice and human cadavers. However, also the samples of the internal organs showed a bi-phasic detection pattern over the course of time. The second group represents gnotobiotic mice with a complex human microbiome. This sample group showed a similar detection pattern as the mice with a conventional murine microbiome (17).

The study of Metcalf et al. (1) from the year 2013 studied samples of body sites and associated soil samples as well as control samples of the soil. During the studied time frame, they observed a change in the microbial communities after the rupture of the abdomen which leads to the decrease of anaerobic taxa and to an increase of aerobic taxa due to the oxygen exposure. The group conclude that this shift in presence and relative abundance of specific taxa could be used to estimate the PMI. For this reason, the group applied a machine learning approach and they were able to create a prediction model which could estimate the PMI within 3.30 ± 2.52 days by using the changes in the microbial communities of the skin cadaver heads. The observed shift from anaerobic to aerobic taxa was expected to be a useful prediction marker for the PMI estimation (1). However,

(15)

this shift from anaerobic to aerobic is contradictory to other studies (5,7,9,12) which observed a shift in the opposite direction. The shift from aerobic to anaerobic taxa was already proposed by Evans in 1963 (21).

The first study of Pechal et al. (6) from the year 2013 was divided into 2 parts. The first part studied the variation of the microbial communities within the 4 seasons, while the second part studied the annual variation between the years 2010 and 2011. Additionally, the influence of insect access and exclusion was investigated. The results showed location-dependent microbial activity which was also observed in other studies (5,18-19). Furthermore, a seasonal and annual variation was found in the microbial communities which could possibly be influenced by the capability of different insects to access the cadaver and deposit their eggs (6). During the winter season, insect activity is reduced and can lead to a delayed access to the cadaver. This results in the conclusion that necrophagous insects could have an impact on the microbial communities during the whole decomposition process (6).

In the second study of Pechal et al. (18) from the year 2014 the bacterial communities changed in taxon richness and relative abundance over the studied time frame of 5 days. In the studied time frame, the cadavers decomposed rather quickly and reached the dry stage within 5 days of outdoor placement (18). For each stage of decomposition, a unique profile of 4 dominant bacterial phyla (Bacteroidetes, Proteobacteria, Actinobacteria and Firmictues) could be observed as well as a trend in relative abundance. Furthermore, a negative linear relationship of phylum richness was observed within the course of time as well as a difference in bacterial communities on each sampling day and each sampling region (18). Additionally, 5 bacterial phyla were identified as crucial predictors of the PMI in the proposed statistical model of the research group. A model with 10 different taxa provided the most informative model and was able to describe 94.4% of the variations (18).

The study of Chun et al. (19) observed a location-specific difference in bacterial communities. This was also discovered in other studies (5-6,18). However, it is notable that the bacterium Proteus

mirabilis was detected throughout the whole decomposition process at each sampling point (19). As

already describe in the study by Pechal et al. (18), also in this study, different combinations of phyla were dominant at a specific stage of the decomposition process. The first 4 hours, for example, were dominate by Firmicutes while at the latest sampling points the soil-associated Actinobacteria were present as well (19). However, as the study used a culture-dependent method, the validity of the results is limited to culturable and aerobic bacteria (19).

Burcham et al. (7) chose a different approach of infecting mice with S.aureus and following the transmigration with whole-body fluorescence imaging. With this technique is was possible to detect an increased spreading of S.aureus over time and with progressive decomposition (7). The transmigration of S.aureus in infected, non-surface sterilised mice occurred at an earlier point in time than in infected, surface-sterilised mice (7). A possible reason for this is the natural occurrence of S.aureus on the skin surface of non-surface sterilised mice. After 30 days, an increased auto-fluorescence was discovered which leads to the conclusion that the transmigration of S.aureus in infected mice cannot be reliably followed after this time point. Additionally, organs were analysed for cultural growth. The different organs showed undulating positive results over the first 14 days after death. After this period, the samples were culture-negative (7). Furthermore, Burcham et al. (7) observed a distinct shift from aerobic to anaerobic taxa throughout the decomposition process, which was also observed by other study groups (5,9,12). In addition, a variation between the replicate mice cadavers was observed, which could influence the reproducibility of results (7). Handke et al. (20) was the only group did not use a whole animal model but reduced their samples to pieces of pork loin. This seems to have an influence on the decomposition process itself as the samples did not decompose as expected but rather dried out (20). An explanation for this could be the sterile conditions in which the samples were kept as well as lack of intestinal bacteria. The

(16)

16 research group also investigated the possibility of a new and more cost-efficient approach than NGS for the identification of bacterial communities. The chosen T-RFLPs technique was not capable to identify up to 50% of the abundant bacterial communities (20). Because of this, it is difficult to assess the true pattern of occurrence of specific bacterial communities within the course of time and within different stages of decomposition.

Additionally, to the human cadaver study, Metcalf et al. (15) used a mouse model which decomposed under natural conditions. The research group investigated the influence of the soil type on the decomposition process. With a prediction model of 1 soil type, they were able to estimate the PMI of a cadaver placed on another soil type (15). Furthermore, they considered the soil to be a possible source of the microbial communities which are involved in the decomposition process. However, the type of soil was not considered to have a major influence on the process of decomposition (15). Metcalf et al. (15) did also recognise a convergence of microbial communities of the cadaver and the microbial communities of the surrounding soil during the process of decomposition (15).

2.3. Summary

In the following part, a short summary of the analysed papers as well as the differences and similarities between these papers are presented. The microbial communities within an animal as well as within a human cadaver experience a change over time as described by several studies (4,7,10-11,15). During the decomposition process, a distinctive shift from aerobic to anaerobic bacterial communities was observed (5,7,9,12), however also a shift from anaerobic to aerobic bacterial communities was observed by Metcalf et al. (1), in contradiction to the other studies and the proposition of Evans in 1963 (21). It was also observed that the bacterial communities of the later decomposition stages more and more align with the bacterial communities of the surrounding soil (5,14-15) which is supported by the presence of Actinobacteria, a soil-associated bacterial phylum (6,14,18-19). Furthermore, the different studies observed a location-dependent microbial activity (5-6,18-19), seasonal and annual variation (6), replicate variation between different cadavers (5,7,11), a difference between the microbiome on the right and left side of a cadaver (11), sex-specific difference in human cadavers (8,12) as well as different microbial communities depending on the cause of death (12). Nonetheless, 5 study groups were able to establish a prediction model for the estimation of the PMI (1,3-4,15,18). The models predicted the PMI with a mean average error of 55 ADD (3), up to a PMI of 600 CDH (4), across seasons (15) or within 3.30 ± 2.52 days (1). This would be a significant improvement to currently used PMI estimation methods. Another interesting observation is the presence of Ignatzschineria, which is associated with fly larvae, in several samples (10-11,16). This raises the question of the influence of necrophagous insects on the decomposition process. The study of Pechal et al. (6) also observed a possible effect of necrophagous insects on the decomposition process and the possible interactions between blow flies and the microbial communities. Additionally, a difference in the microbiomes was observed when extrinsic factors (10) and adipocere formation (14) influence the decomposition process. This raises the question of the impact of mummification on the microbial communities. Furthermore, several studies detected a unique profile and trend in relative abundance for each stage of decomposition (15,18-19). This could help to establish a method for a more precise estimation of the PMI, since one or more microbial profiles could be assigned to a specific stage of decomposition and subsequently a specific time frame. However, also a bi-phasic, undulating detection pattern was observed (7,17), which could lead to a more complex interpretation of the results and therefore a need for further research. The following table 4 summarises the results.

(17)

Table 4 Summary of the results

Summary of the results

Shift from aerobic to anaerobic bacteria (5,7,9,12)

Alignment of human microbiome and bacterial communities of the soil (5-6,14-15,18-19) Location-dependent microbial activity (5-6,18-19)

Seasonal and annual variation (6) Replicate variation (5,7,11) Sex-specific differences (8,12)

Possible influence of necrophagous insects on the decomposition process (6,10-11,16)

Unique microbial profile and trend in relative abundance for each stage of decomposition (7,15,19)

Clostridium as potential prediction marker for PMI (5,8-12,16-17)

In table 5 an overview of the mentioned bacteria in the different studies is presented. Remarkably is the presence of the phylum Firmicutes in nearly all the studies (1,4-13,16-18). It also needs to be mentioned that the bacterium Clostridium is present in 8 of the 19 studies (5,8-12,16-17). In several human cadaver studies, Clostridium was observed in a large number (5) and throughout the whole decomposition process (12). Due to this it was proposed that Clostridium could be a stable biomarker (8) and good prediction marker (8,10) for the PMI.

(18)

Table 5 Overview of the mentioned bacterial phyla in the different studies. The green colour symbolises the presence of the phylum in the corresponding study. Additionally, important bacteria are mentioned in the column of the corresponding phylum.

Proteobacteria Firmicutes Actinobacteria Bacteroidetes Acidobacteria Tenericutes Fusobacteria Adserias-Garriga

et al. (16) Ignatzschineria Clostridiales, Bacillaceae

Burcham et al. (7)

S.aureus, C.perfringens (limited to these bacteria)

Can et al. (9) Clostridium, Lactobacillus

Chun et al. (19) Proteus

Damann et al. (14) Alpha, Gamma soil soil

DeBruyn and Hauther (10)

Gamma

(Ignatzschineria) Clostridium

Handke et al. (20)

Hauther et al. (4) Lactobacillus Bacteroides

Heimesaat et al.

(17) Clostridium

Prevotella, Bacteroides

Hyde et al. (5) Clostridium

Hyde et al. (11) Actinetobacter,

Ignatzschineria Clostridium

Javan et al. (8) Clostridium Prevotella

Javan et al. (12) Clostridium

Johnson et al. (3) Metcalf et al. (15)

Metcalf et al. (1) Alpha, Gamma bloat stage bloat stage

Pechal et al. (6) Pechal et al. (18) first sampling day Tuomisto et al. (13) Staphylococcus, Streptococcus

(19)

3. Discussion of literature

3.1. Differences between human cadaver and animal models

Comparing the found literature, it is necessary to differentiate between a human cadaver and an animal model. There are different views on the usability of animal models. In general, animal models such as pigs and mice are frequently used in the medical field to test drugs or use parts such as the heart valve to replace failing human parts. Furthermore, pigs are sometimes used to model the human body. However, Hyde et al. (5) stated in their first paper from 2013 that they consider the pig microbiome to be not comparable to the human, which would have a significant impact on the animal cadaver studies. One way to overcome the possible limitation of animal models was demonstrated by Heimesaat et al. (17), who created gnotobiotic mice with a complex human microbiome. Nonetheless, this is not the ideal situation to investigate the human microbiome and the post-mortem changes of it. Finally, it is necessary to study the human microbiome with human cadavers. A major disadvantage of human cadavers lies in the impossibility to replicate and repeat the results since every human has a very distinct microbiome, which is influenced by many extrinsic and intrinsic factors.

3.2. Influence of extrinsic and intrinsic factors

Extrinsic and intrinsic factors also have an influence on the chosen sample site of the body. External sample sites, like the skin of the cadaver or the oral cavity are much more influenced by environmental factors than internal sample sites, like the gut or other internal organs which are supposed to be sterile in the living human (13). The influence of these factors is also visible when looking at the microbial communities of different body sites and between different bodies. Several study groups identified a location-dependent microbial community (5-6,18-19) and a difference between cadavers (5,7,11). However, both differences are reasonable but need to be considered in the research. When dealing with human cadavers, it is hardly possible to have replicate samples since every human being bears a distinct microbiome which is influenced by the daily live and many other factors. Also, the body site specific differences are reasonable since each body site or organ has a specific function which needs a specific microbiome or the exclusion of bacteria from the organ.

3.3. Influence of the sampling strategy

When choosing a sample site for a study, it is important to consider the whole sampling strategy. Is it necessary to access the sampling site without disturbing the body because multiple samples must be taken at multiple sampling points from the same area? In this case, an external sample site is preferably chosen. Is a site which is less affected by the environment considered and is only one sampling point enough? In that case, an internal sampling site is suitable for the study. However, it is necessary to be aware of the possible influencing factors on each sampling site which could be part of future research. Additionally, with respect to the sample site specific and body specific differences, it is also important to consider possible sex-specific differences which were discussed in the studies by Javan et al. (8,12).

3.4. Additional possible influencing factors which raise questions

Other influencing factors are not extensively studied like for example the influence of the diet of the person or possible undetected and detected diseases. The study of DeBruyn and Hauther (10) considered a feeding tube to be the reason for a distinct different starting microbiome in 1 of the human cadavers. Unfortunately, this body was excluded from the study, which leads to the necessity to conduct further research in this area. Another interesting possibly influencing factor is the use of antibiotics prior to death since these drugs are meant to kill the studied bacteria. The question which arises is, if the reduction of specific bacteria has an extensive impact on the decomposition process itself. Another factor which belongs in this category is the possible influence of the cause of

(20)

20 death on the decomposition process. Javan et al. (12) discussed a possible influence of the cause of death on the decomposition process. In the studies of Hyde et al. (5) and Tuomisto et al. (13) the cause of death ‘poisoning’ was mentioned. A question which arises specifically to this cause of death is if it could possibly alter the microbiome prior to death and therefore cause an altered course of the decomposition process. In the study of Hyde et al. (5), a carbon monoxide poisoning is one cause of death. A carbon monoxide intoxication leads to the enrichment of carbon monoxide in the blood since it has a higher affinity to haemoglobin than oxygen. This could lead to a hypoxia of the tissue. The question which arises is if this resulting ante-mortem hypoxia has an influence on the composition of the post-mortem bacterial communities and the whole decomposition process.

3.5. The shift from aerobic to anaerobic taxa

Evans (21) already described a shift from aerobic to anaerobic taxa during the decomposition process. At first instance, aerobic bacteria reduce the amount of oxygen in the tissues and create chemical conditions which are suitable of anaerobic bacteria. These anaerobic bacteria than take over at a later point during the decomposition process (21). Several studies observed a distinctive shift from aerobic to anaerobic taxa during the decomposition process (5,7,9,12) which could possibly be used for an estimation of the PMI. Nonetheless it needs to be established how reproducible this shift is in humans. In relation to the before mentioned poisoning as cause of death, the question is if this shift in the microbial communities could occur at an earlier time point since the oxygen concentration in the tissue is already lower than in the usual case. This factor needs to be considered as a possible limitation in the estimation of the PMI with the microbiome.

3.6. Influence of temperature on seasonal and annual variation

Another extrinsic factor, which plays a key role in the case of outdoor placement above the ground, is the temperature change throughout the different seasons and the difference in temperature between years. Pechal et al. (6) did an extensive study on this topic with an animal model and observed a seasonal and annual variation. One way to account for the daily differences in temperature is the use of accumulated degree days or cumulative degree hours for a better temperature-time comparison. However, 1 study is not enough to verify these results and additional research is necessary.

3.7. Different decomposition patterns

In relation to the outdoor conditions and the temperature, the processes of adipocere formation and mummification need to be mentioned. These processes occur in environments with high moisture in the first case or very dry or cold regions in the second case. In the case of adipocere formation, Damann et al. (14) observed different results for the cadaver with adipocere formation than results from samples with a similar PMI, which indicates an influence of the adipocere formation on the decomposition process. For the PMI estimation of mummies, a different method such as radiocarbon dating could be performed. But also during the decomposition under normal conditions, the bodies can decompose in a different pattern which can lead to difficulties in consistent sampling throughout the whole sample size. This is especially considerable when in one body a specific part is not any more recognisable and cannot be sampled and in another body the part can still be sampled. This has an influence on the later results.

3.8. Influence of insects and soil bacteria

Another factor which is noteworthy, is the possible influence of insects and soil bacteria on the decomposition process. The influence of insects can be seen in the occurrence of the bacterium

Ignatzschineria which is associated with flies and was detected by several studies (10-11,16). It can

easily be investigated with an insect-exclusion study. Furthermore, an influence of soil bacteria on the decomposition process was observed. There are quite a few papers published in this area, but they are not considered in this review since the focus lies on the human microbiome. However, this

(21)

is an interesting topic for future research and definitely is worth to be considered. Nonetheless, some studied papers also mention a relation between the decomposition process and the soil bacterial communities. Especially during the later stage of decomposition, the bacterial communities of the cadaver and the soil align with each other and become more similar (11,14-15).

3.9. Influence of the used methodology

Last but not least, the technical part of the study is also capable of influencing the results. The influence of sampling techniques on the results is discussed rather controversially. Hyde et al. (5) reported clearly varying results within one body when using 2 different sampling techniques for the same sampling site. Hauther et al. (4) also indicated a possible influence of their sampling method on the results. However, on the other hand, DeBruyn and Hauther (10) could not find an influence of the sampling technique on their results. Since there are different statements, it is necessary to further evaluate the influence of sampling techniques in a dedicated evaluation study, but in combination with appropriate DNA extraction techniques since this can also influence results. The evaluation of the sampling techniques remains a challenge for the future, since a suitable test surface is necessary with a desirable reproducible microbiome on several different test cadavers. Another point which needs to be considered in this category, is the chosen analysis method. There are 2 approaches, namely the culture-based approach and the culture-independent approach. Advantages of the based approach are that it is much cheaper than NGS. However, culture-based techniques also have a crucial disadvantage compared to culture-independent techniques since they only display cultivable bacteria which are estimated to be only around 1% of the bacteria found in nature (5).

(22)

22

4. Conclusion and recommendations

Conclusion

This review focused on the present literature studying the evolution of the human microbiome during the decomposition process, preferably under natural conditions above the ground. For this purpose, studies which used human cadavers and animal models were analysed. Currently the decomposition process itself is not extensively studied. Therefore, animal models can be used in the first instance to better understand this decomposition process. However, animal models are not able to mimic the human microbiome in all its details and therefore human cadavers should be used finally.

Multiple studies observed a change in the microbiome over time and a distinct shift from aerobic to anaerobic bacteria. This reinforces the assumption that the human microbiome changes during the decomposition process which subsequently could be used for an estimation of the PMI. Additionally, the presence of the bacterium Clostridium was observed in 8 of the 19 studies which indicates that

Clostridium could be a stable bio-and prediction marker for the PMI estimation.

The questions which remain open concern the many influencing factors which were addressed in some studies but need further evaluation. These factors are for example the reproducibility of the results when human cadavers are used, the influence of necrophagous insects and soil bacterial communities and the influence of the sampling and extraction methods.

Recommendations

Generally, it can be said that much more research is necessary to answer all these questions and be able to adequately account for the influencing factors to accurately estimate the PMI. Future studies should consider a larger sample size and clearly defined characteristics of the bodies, e.g. a known PMI, known medical history and known cause of death. Prior to the placement of the bodies, the sampling and extractions methods should be clearly defined to avoid possible influences during the later examination. This review only focused on bodies placed above ground. Future research should also consider other forensically relevant body placements like for example buried or immersed bodies which undergo a possibly different decomposition process with different microbial communities involved. There is already some literature about the influences of the soil bacterial communities which should also be considered in future research.

In the end, it may be useful to combine several internal and external sampling sites to avoid site-specific irregularities and subsequently combine different estimation methods to increase accuracy of PIM estimation. Concluding, a lot of factors need to be considered when using the microbiome to determine the PMI. Much more research needs to be done to answer most of the relevant questions to be able to accurately predict the PMI with the microbiome in the future. However, the review of the present available literature showed already some promising approaches.

(23)

5. References

1. Metcalf J, Wegener Parfrey L, Gonzalez A et al. A microbial clock provides an accurate estimate of the postmortem interval in a mouse model system. eLife. 2013;2

2. Madea B. Methods for determining time of death. Forensic Science, Medicine, and Pathology. 2016;12(4):451-485.

3. Johnson H, Trinidad D, Guzman S, Khan Z, Parziale J, DeBruyn J et al. A Machine Learning Approach for Using the Postmortem Skin Microbiome to Estimate the Postmortem Interval. PLOS ONE. 2016;11(12):e0167370.

4. Hauther K, Cobaugh K, Jantz L, Sparer T, DeBruyn J. Estimating Time Since Death from Postmortem Human Gut Microbial Communities. Journal of Forensic Sciences. 2015;60(5):1234-1240.

5. Hyde E, Haarmann D, Lynne A, Bucheli S, Petrosino J. The Living Dead: Bacterial Community Structure of a Cadaver at the Onset and End of the Bloat Stage of Decomposition. PLoS ONE. 2013;8(10):e77733.

6. Pechal J, Crippen T, Tarone A, Lewis A, Tomberlin J, Benbow M. Microbial Community Functional Change during Vertebrate Carrion Decomposition. PLoS ONE. 2013;8(11):e79035.

7. Burcham Z, Hood J, Pechal J et al. Fluorescently labeled bacteria provide insight on post-mortem microbial transmigration. Forensic Science International. 2016;264:63-69.

8. Javan G, Finley S, Can I, Wilkinson J, Hanson J, Tarone A. Human Thanatomicrobiome Succession and Time Since Death. Scientific Reports. 2016;6(1).

9. Can I, Javan G, Pozhitkov A, Noble P. Distinctive thanatomicrobiome signatures found in the blood and internal organs of humans. Journal of Microbiological Methods. 2014;106:1-7.

10. DeBruyn J, Hauther K. Postmortem succession of gut microbial communities in deceased human subjects. PeerJ. 2017;5:e3437.

11. Hyde E, Haarmann D, Petrosino J, Lynne A, Bucheli S. Initial insights into bacterial succession during human decomposition. International Journal of Legal Medicine. 2015;129(3):661-671.

12. Javan G, Finley S, Smith T, Miller J, Wilkinson J. Cadaver Thanatomicrobiome Signatures: The Ubiquitous Nature of Clostridium Species in Human Decomposition. Frontiers in Microbiology. 2017;8.

13. Tuomisto S, Karhunen P, Vuento R, Aittoniemi J, Pessi T. Evaluation of Postmortem Bacterial Migration Using Culturing and Real-Time Quantitative PCR. Journal of Forensic Sciences. 2013;58(4):910-916.

14. Damann F, Williams D, Layton A. Potential Use of Bacterial Community Succession in Decaying Human Bone for Estimating Postmortem Interval,,. Journal of Forensic Sciences. 2015;60(4):844-850. 15. Metcalf J, Xu Z, Weiss S et al. Microbial community assembly and metabolic function during mammalian corpse decomposition. Science. 2016;351(6269):158-162.

16. Adserias-Garriga J, Quijada N, Hernandez M, Rodríguez Lázaro D, Steadman D, Garcia-Gil L. Dynamics of the oral microbiota as a tool to estimate time since death. Molecular Oral Microbiology. 2017;.

17. Heimesaat M, Boelke S, Fischer A et al. Comprehensive Postmortem Analyses of Intestinal Microbiota Changes and Bacterial Translocation in Human Flora Associated Mice. PLoS ONE. 2012;7(7):e40758.

(24)

24 18. Pechal J, Crippen T, Benbow M, Tarone A, Dowd S, Tomberlin J. The potential use of bacterial community succession in forensics as described by high throughput metagenomic sequencing. International Journal of Legal Medicine. 2014;128(1):193-205.

19. Chun L, Miguel M, Junkins E, Forbes S, Carter D. An initial investigation into the ecology of culturable aerobic postmortem bacteria. Science & Justice. 2015;55(6):394-401.

20. Handke J, Procopio N, Buckley M et al. Successive bacterial colonisation of pork and its implications for forensic investigations. Forensic Science International. 2017;281:1-8.

(25)

Appendix

The appendix contains the applied search strategy and the found papers which have a relation to the PMI and/or the human microbiome.

Search on lib.uva.nl

- Search term: “PMI estimation microbiome” - Results: 22 hits

- Scan titles and abstracts for interesting papers - Found papers

o Adserias-Garriga et al., 2017a o Clarke et al., 2017

o Brooks, 2016 o Javan et al., 2016b o Chun et al., 2015 o Guo et al., 2016

o DeBruyn and Hauther, 2017 o Finley et al., 2015

o Finley et al., 2014

o Castillo-Peinado and Luque de Castro, 2016 o Handke et al., 2017

o Iancu et al., 2015 o Olakanye et al., 2017

- Search term: “postmortem microbiology forensics” - Results: 48 hits

o Limit to ‘articles only’ results in 44 hits - Scan titles and abstracts for interesting papers - Found papers which were not already in the list

o Burcham et al., 2016 o Carter et al., 2015 o Javan et al., 2017 o Hyde et al., 2015 o Vass et al., 1992

(26)

26 Search on google scholar

- Search terms: “Thanatomicrobiome” - Results: 68 hits

- Scan of abstracts for interesting papers

- Found papers with interesting title not related to medical issues like sepsis and not already found beforehand

o Can et al., 2014 o Javan et al., 2016a o Thomas et al., 2017 o Tuomisto et al., 2013 o Javan et al., 2015 o Benbow et al., 2015

o Adserias-Garriga et al., 2017b o Finley et al., 2016a

o Finley et al., 2016b o Maile et al., 2017

o Noble and Pozhitkov, 2017 o Pascual et al., 2017

o Thomas et al., 2017

Check references of Javan et al., 2016b for interesting articles - Benninger et al., 2008

- Carter et al., 2008a - Carter et al., 2010 - Damann et al., 2015 - Dickson et al., 2011 - Hauther et al., 2015 - Heimesaat et al., 2012 - Howard et al., 2010 - Hyde et al., 2013 - Metcalf et al., 2013 - Metcalf et al., 2016a

(27)

- Pechal et al., 2014 - Pechal et al., 2013

- The human microbiome project consortium, 2012

Check references of DeBruyn and Hauter, 2017 for interesting articles - Cobaugh et al., 2015

- Johnson et al., 2016 - Mathur and Agrawal, 2011

Adserias-Garriga et al., 2017b – References - Schloissnig et al., 2013

- Finley et al., 2014

Adserias-Garriga et al., 2017a – References - No additional papers

Mathur and Agrawal, 2011 – References - No additional papers

Carter et al., 2015 – References - Vass, 2011

- Meyer et al., 2013 Hauther et al., 2015 – References

- Carter et al., 2008b - Costello et al., 2009

- Jalanka-Tuovinen et al., 2011 Hyde et al., 2015 – References

- Vass, 2001

- Butzbach et al., 2013 Handke et al., 2017 – References

- Pittner et al., 2017 - Graf et al., 2015 - Stokes et al., 2013 Javan et al., 2016a – References

- Turnbaugh et al., 2007 Search on Web of Science:

New approaches in estimating the post-mortem interval (PMI) – The human microbiome as an indicator of the time since death