Time estimation of a biological trace found at a crime scene represents one of the most important aspects in forensic casework, together with identification of the trace donor by DNA profiling. Knowledge about trace timing allows decisions to be made regarding whether the identified donor could potentially be the perpetrator and may also help to find potential eye witnesses or additional suspects. In principle, time aspects of crime scene samples can be seen from two sides. One side addresses the question of the time interval a sample was present at the crime scene, which may be addressable by differential degradation of biomarkers, such as different kinds of RNAs, but more detailed studies on this issue are needed [1]. The other side, which we address in this study, concentrates on determining the time point (i.e. the time of day) a trace was deposited at a crime scene, thereby bringing a chronobiological aspect into forensic research.

The 24-h day–night rhythm has always been the most reliable environmental cue for life on earth. Behavioural characteristics, metabolism and body functions are adjusted constantly to changing light and dark periods [2]. One of the best studied biomarkers of the human circadian pacemaker is the hormone melatonin, which signals to the body the duration of the dark period [3]. Melatonin is only produced and secreted during the night and shows its peak concentration in the later part of the night [3]. Cortisol is also widely used as a biological marker of circadian phase, often in parallel with melatonin [2, 4]. Cortisol levels display peak concentrations in the morning shortly after wakening and decrease throughout the day and night [5].

In this study, we investigated the feasibility of using circadian hormones, such as melatonin and cortisol, for estimating trace deposition time. We employed commercially available enzyme-linked immunosorbent assays (ELISA) easily applicable in forensic laboratories. To our knowledge, this is the first study exemplifying that deposition time of forensic traces can be estimated by means of circadian biomarkers, although the use of melatonin for time of death estimation has been described before [6, 7].

With this study, we demonstrate for the first time the feasibility of estimating deposition time of biological traces by means of circadian biomarkers. Our approach includes determination of melatonin and cortisol concentrations and their normalization against total protein concentration to control for variation in sample input. This biochemical tool is expected to enable a forensic researcher to narrow down the time of deposition of a trace sample collected at a crime scene to either late night (melatonin) or early morning (cortisol). We chose a simple method based on ELISA because (1) it allows the use of commercially available assays that are fully established in clinical diagnostics including quality control, (2) sample processing is straight-forward and (3) only basic laboratory equipment is needed. Alternatives for hormone quantifications include chromatographic methods such as GC/MS or LC/MS. Studies of human circadian rhythms aiming to establish biological relationships normally involve strictly controlled laboratory conditions (i.e. control of lighting levels, food intake, posture, sleep etc.) for sampling, including also the so-called constant routine protocol, which is used to unmask endogenous rhythms from any confounding factors [4, 11, 12]. However, such controlled conditions were not applied in our study aiming to provide a tool for future forensic applications, and it is highly remarkable that we were still able to reproduce both hormone profiles with high accuracy. This provides encouraging evidence that our approach can be transferred to forensic practice.

One key aspect of the ELISA assay applicability testing was downscaling of the sample volumes necessary for the assays to meet requirements of forensic trace analysis. With the almost perfect correlation of the values obtained for the same blood and serum samples using two different cortisol ELISA kits, we were able to achieve this goal with only 1 μl of sample (instead of 20 μl) needed. In addition, in order to use whole blood samples with the melatonin saliva kit employed in this study, melatonin was extracted from the samples to prevent matrix problems and false measurements before applying the samples to the ELISA plate. This extraction procedure has already been proven as feasible approach using homogenized tissue of post-mortem human pineal glands [6]. Overall, a total volume of 25 µl of blood, corresponding to a cotton blood stain size of only 70 mm², is required to perform all melatonin, cortisol and total protein measurements in duplicate. A minimal cotton blood sample size of 50 mm² would be required if single measurements were employed. In contrast, a much higher total volume of 160 μl saliva was needed for all measurements in duplicates.

The highly significant correlations between corresponding values for saliva and blood for both melatonin and cortisol are in agreement with previously published data comparing hormone concentrations between blood, serum, and saliva [13–16]. The observation that cortisol concentrations obtained from whole blood reached only 67% on average when compared with the corresponding serum concentrations may be explained by the dilution of the serum component with the rest of the blood compounds present in whole blood, leaving the same amount of cortisol in a larger volume for the assay. However, this effect was not seen with melatonin, where similar values from blood and serum were obtained. It has been shown recently that melatonin is also produced by immune cells present in the blood, providing a melatonin pool that may not be accessible in serum alone [17].

The 24 h profiling for total protein concentration and hormone levels could be successfully performed for blood samples. The finding of a remarkably stable total protein concentration during the entire day–night period, both within and between subjects, is in agreement with previous data on serum and cellular blood components [18]. Hence, these data allow the conclusion that normalization of hormone levels against total protein levels can be considered a valid approach. Both melatonin and cortisol profiles showed the typical 24 h pattern, with subjects who were employed in different parts of this study (validation, profiling, degradation) exhibiting high reproducibility [4, 10–12, 19]. Comparison of our calculated mean hormone patterns with the respective reference curves also confirmed the previous statement that melatonin values in blood reach similar levels to those in serum, whereas values for cortisol are found to be lower [10].

Our preliminary blood stain analysis revealed a good transferability of the methods established on fresh blood. Surprisingly, the recovery rate of total protein and cortisol came very close to the reference tubes from freshly frozen samples, independent of the surface material. For melatonin, the recovery rate was lower (although not statistically significant). As expected, resolubilisation of dried blood was achieved more easily from non-porous surfaces. When testing blood samples stored up to 4 weeks, total protein concentrations were in general surprisingly stable across the entire time period within and between individuals. As such, we conclude that the use of this parameter for normalization is a valid approach for forensic samples, at least within the storage time limit of 4 weeks tested in our study. This conclusion was strengthened by the highly significant correlations found for measured against normalized values for both hormones. In our hands, melatonin concentrations from blood samples did not decrease significantly with sample storage time of 4 weeks after normalization, which is in line with previous data on post-mortem melatonin degradation [7]. Interestingly, there was no remarkable difference observed between the sample set kept in the dark and the one kept near the window, suggesting that, at least under these experimental conditions, photodegradation of melatonin is negligible [20]. Although there is no report on melatonin stability in stains thus far, it should be mentioned here that melatonin is reported to be photolabile in solution [20]. On the other hand, cortisol levels from blood declined significantly with sample storage time, which is in contrast to previous reports on the stability of cortisol in serum and whole blood, yet to our knowledge there is no study on the stability of cortisol in blood stains on cotton or parafilm in the literature [21, 22]. Therefore, the observed time-wise decay suggests that the use of this hormone may be difficult in samples stored for longer time intervals. We would advise the implementation of a strict threshold value for use and interpretation of the result. In our albeit limited study, the cotton sample set would be used as a reference and lead to a threshold value of 2.5 × 10⁻⁷ (cortisol/protein normalized), or 7 days degradation time, below which no more conclusions can be drawn on the sample deposition time. However, we would like to emphasize that more detailed forensic validation studies are needed to establish more stringent regulations for this issue.

One interesting question raised by our study is whether the same stain sample can be used for multiple purposes. To test this, we used part of the melatonin extraction column flow-through from two 20 µl fresh blood samples for DNA extractions applying the QIAamp DNA Micro extraction system (Qiagen), and were able to generate complete DNA profiles with the AmpFlSTR Identifiler PCR amplification kit (Applied Biosystems). In addition, by using another part of the same flow-through it was possible to extract total RNA after phenol extraction with the RNeasy Micro Kit (Qiagen), to synthesise cDNA with the RevertAid H Minus First Strand cDNA Synthesis Kit (Fermentas), and to successfully perform mRNA profiling for blood specific markers (data not shown) [23].

In general, the situation in saliva turned out to be more difficult compared to whole blood, providing less consistent and firm results. The large differences observed in total protein concentrations within and between individuals confirm previous findings, and do not allow recommending this parameter for normalization [18, 24]. Also, the amount of saliva needed for the assays might affect saliva usability in forensic trace analysis. High-viscosity saliva resulted in falsely increased melatonin values, and for some of the subjects the peak in total protein concentration fell together with the melatonin peak in the middle of the night, cancelling out each other in normalization. It has been described before that protein degradation in saliva is extremely rapid compared to other biological fluids, and starts immediately upon sample collection [25]. The dynamic nature of the saliva proteome shows high variation in protein concentration between and within individuals [18, 24]. Thus, it remains to be shown if highly stable protein markers with constant 24 h expression can be established. Until such markers for normalization of variation in saliva input are identified, the use of saliva in forensic applications of circadian biomarkers is discouraged.

It shall be mentioned that the use of the circadian hormones melatonin and cortisol in future forensic applications is not without limitations. One aspect is that some individuals represent extreme chronotypes (‘owls’ or ‘larks’; approximately 2% of the population, not part of our study) with significantly shifted profiles in their circadian rhythms. Furthermore, night-time melatonin biosynthesis is suppressible by exposure to light [3, 26, 27]. It has been shown that a three hour night-time light exposure (600 lux) decreased melatonin concentrations by 65%, whereas cortisol levels were not significantly altered [28]. Furthermore, an association between reduced melatonin levels with depression is reported in the literature but discussed controversially [29, 30]. Cortisol, besides its role as a circadian marker, is also known as biomarker for stress. Although it is not immediately up-regulated in acute (psychological) stress situations but instead exhibits a delayed response, implying little or no influence on crime scene samples [31, 32]. Yet, several human and animal studies suggest associations between altered cortisol levels and physical or mental symptoms, like posttraumatic stress disorder or addiction (see [5, 33] and references therein). Another aspect is that our approach is based on the fact that every human being is entrained to the environment, mainly by the synchronizing factor light. Yet there are a few exceptions from this general principle, like blind persons who have no conscious light perception and therefore may show shifted circadian patterns [34]. Disturbed circadian rhythms can also be found for some days after transmeridian flights (‘jet-lag’) and in shift workers with slowly rotating schedules (see [35], and references therein). Furthermore, reduced amplitudes in circadian rhythms, including the melatonin rhythm, have been described in the elderly, although this phenomenon needs to be studied more systematically [36, 37]. Thus, data obtained for both melatonin and cortisol in future forensic applications need to be interpreted carefully and in the context of available information on the case background.

Despite of the potential caveats listed we believe that the presented approach based on the circadian hormones melatonin and cortisol bears large potentials for a first estimation of the time a biological trace was deposited at a crime scene, which can provide crucial information for solving crimes. Systematic studies in search for more circadian biomarkers shall be carried out in the future to see if the time range offered by the two hormones studied here can be specified more exactly. In addition, our approach may be transferable to forensic cases involving whole bodies rather than traces: Pineal melatonin content has been suggested as an additional indicator for estimating time of death [7], yet confounding factors such as depression/suicide have to be considered [38]. Also, melatonin levels in different body sites were found to be not significantly related to post-mortem interval or gender, but highly correlated with each other [16]. Still, future research is needed to confirm the feasibility of such an approach. Besides the anticipated forensic application, there may also be clinical applications for the described methods. The small blood volume needed to perform the circadian hormone analyses could be beneficial, for instance in the monitoring of disturbed melatonin production and/or its supplementation treatment in premature infants, or taking advantage of the easily performed sampling via finger-prick, which allows for multiple sampling even in children or in elderly people in a domestic environment, a situation often found in field studies on ageing or biological rhythm disorders [39–42].