Abstract
The time-dependent postmortem increase of potassium concentration in the eye fluids has been studied since the 1960s. However, important discrepancies on the reproducibility of the phenomenon have hampered the use of this parameter in real cases. In recent years, a new analytical approach based on capillary ion analysis (CIA) has been reported. In the present work, the correlation between vitreous potassium and postmortem interval (PMI) has been re-evaluated by using CIA in a group of 164 cases with PMIs ranging from 2 to 110 hours. The correlation of the two parameters was described by the following regression equation: y = 0.1733x + 2.3008 (x = PMI; y = K+ concentration); correlation coefficient = 0.962. The re-calculation of PMIs on the basis of this equation provided calculated PMIs with an average error of 5.54 hours (SD = 4.16). However, the percent PMI calculation error decreased with the increase of PMI, becoming acceptable for practical application above 24 hours since death.
Introduction
In the 1960s studies looked at the use of the potassium concentration in the vitreous humour as a potential tool for the determination of postmortem interval (PMI). 1 The promise of an independent test for estimating time of death is tantalizing, and a large body of literature has accumulated, which has highlighted, on the one hand, the time-dependent postmortem increase of potassium concentration in the eye fluids, but on the other, important discrepancies among the different authors about the characteristics and the reproducibility of the phenomenon. In particular, the equations reported by the different authors describing the relationship between PMI and vitreous potassium concentration differ in both slope and intercept of the linear regression lines. Therefore most of the forensic pathology community does not believe that the use of vitreous potassium concentration as an estimator of PMI fully meets the requirements of reliability needed for application to real cases.2,3 To overcome these problems, more sophisticated mathematical approaches such as the loess smooth curve have also been proposed to describe the phenomenon. 4 In addition, some authors have proposed to study multiple vitreous humour components, including calcium, magnesium, sodium, hypoxantine and ammonium, also using multivariate analysis of data.5–11
A critical evaluation of the existing literature suggests, as reported in a recent review of Madea and Musshoff, 3 that the differences mentioned above may be accounted for by taking into consideration the different sampling procedures and the different analytical methods used in the different studies. As a matter of fact, very often potassium analysis has been carried out with non-separation methods used in clinical chemistry (e.g. flame photometry and ion selective electrodes). However, these methods have been validated for plasma or serum where the potassium concentration range is much more limited and the matrix composition is more reproducible in comparison to postmortem fluids.
In the last decade a new analytical approach based on capillary ion analysis (CIA) has been reported, which combines high efficiency electrophoretic separations of ions with highly selective indirect UV detection based on the electrostatic displacement of a charged UV absorbing additive.12,13 This technique proved able to analyse potassium in untreated vitreous humour even in the presence of high concentrations of other cations like sodium, calcium, magnesium and ammonium as well as in highly degraded samples. These characteristics make CIA suitable for the analysis of real samples, overcoming much of the possible specific and non-specific interference, which could affect the results obtained from traditional clinical chemistry techniques. Also, the limited quantity of samples required for CIA (few microlitres) allows the collection of microsamples of vitreous humour, thus limiting the interference caused by the artefactual leak of potassium from retinal cells during suction of the eye fluid. CIA was also applied to study a reported difference14–16 in potassium concentration between the two eyes. 17 The results clearly showed no statistical difference, suggesting that the reported between-eye discrepancies could be ascribed to analytical imprecision of the used methods. These findings were later confirmed by studies of Gagajewski et al. 9 and Mulla et al. 10 where vitreous analysis was performed by using a method based on ion selective electrodes.
A further potential cause of variability could be found in the extended range of the investigated PMI spanning from a few hours up to some days.
On this basis, the present work was aimed at the reevaluation of the correlation between vitreous potassium and PMI by using CIA in a group of cases with PMI ranging from 2 to 110 hours. Special attention was paid to the comparison of cases with PMI ≤ 20 hours and >20 hours.
Materials and methods
Standards and chemicals
Imidazole (99% pure) was obtained from Sigma (St Louis, MO, USA), 18-crown-6 ether (99% pure) and a-hydroxybutyric acid (HIBA) (99% pure) from Aldrich (Milan, Italy). Standard solutions of K+ and Ba2+ were prepared from AnalaR® salts (Merck, Darmstadt, Germany). K+ standard solutions were checked versus reference standards for clinical chemistry analysers from Boheringer Mannheim (Mannheim, Germany) containing, respectively, 6.0 and 3.5 mmol/L of K+. Water used for the preparation of the buffers and for sample dilution was of HPLC grade (Carlo Erba, Milan, Italy). The electrophoretic buffer was adjusted to pH 4.5 with 1 mol/L acetic acid. Buffer and rinsing solutions were filtered and degassed under vacuum before use with a membrane filter of 0.22 μm (Millipore, Vimodrone, Italy).
Instrumentation and analytical conditions
A PACE MDQ automated capillary electropherograph (Beckman Coulter, Fullerton, CA, USA) equipped with a filter UV absorbance detector was used throughout the present study. Untreated fused-silica capillaries (75 μm ID, 50 cm effective length, 60 cm total length Beckman Coulter) were used. Beckman 32 Karat Workstation was used for instrument control, data acquisition and processing.
Analytical conditions and method validation were described in detail in a previous paper. 13 In brief, the electrophoretic separations were carried out in a pH 4.5 running buffer composed of 5 mmol/L imidazole, 5 mmol/L 18-crown-6 ether and 6 mmol/L HIBA. Electrophoretic runs were performed by applying a constant voltage of 500 V/cm at 25°C. Detection was by indirect UV absorption at 214 nm wavelength. The samples were injected by nitrogen pressure (0.5 psi) application for 10 seconds at the anodic end of the capillary. Between two consecutive runs, the capillary was washed with water (1 minute) and then with the electrolyte buffer (4 minutes) by applying a pressure of 20 psi at the capillary inlet.
Quantification was carried out on the basis of peak areas by using the internal standard method (barium).
Statistical analysis of data was done by using parametric descriptive statistics and the Student's t-test.
Sample collection and preparation
Vitreous humour samples were collected from 164 medicolegal autopsies of cases of natural or violent deaths, in which the time of death was exactly known (PMI range 1.8-109 hours) either from Jefferson County, Alabama, USA (n = 92) or from the province of Verona, Italy (n = 72). The subjects studied ranged from 11 to 88 years in age, and the causes of death included myocardial infarction, pulmonary embolism, traffic accidents, drowning, gunshot injuries, hanging, precipitation, drug overdosing, and sharp or blunt force injuries. Cases with conditions which could have clearly affected potassium concentrations in biological fluids such as eye diseases, renal failure, dehydration, etc. were excluded. No other preselection of cases was carried out.
The samples of vitreous humour of about 100 μL each were collected from both the eyes by needle puncture (25 G) and gentle suction from the posterior chamber with a plastic syringe (insulin type).
Samples were stored in plastic vials frozen at — 20° C until analysis. Samples collected in the USA were stored in a freezer immediately following collection and subsequently shipped in dry ice by an overnight forwarding agency.
Prior to analysis all samples were diluted 1:20 with a 40 μg/mL aqueous solution of barium chloride, the used internal standard (IS), and then injected.
Results and discussion
The capillary electrophoretic analysis of vitreous humour produced neat electropherograms with no sample preparation other than dilution 1:20. Figure 1 shows a typical electropherogram in which the peaks of ammonium, potassium, sodium and IS (barium) are baseline resolved, notwithstanding huge differences in concentrations between sodium and the other analytes. The analytes migrated in symmetrical peaks with a high efficiency (n = 40,000 for potassium peak). For details of method validation readers are referred to the previous papers of our group.13,17
Typical electropherogram of vitreous humour in which the image is inverted, as is usual in CIA. For the analytical conditions, see the text. CIA, capillary ion analysis
The CIA analysis of the vitreous samples from the 164 cases showed a potassium concentration, expressed as the mean of two eyes, ranging from 2.86 to 23 mmol/L. The plot of PMI with vitreous potassium concentration resulted in a high correlation (Figure 2). This relationship is described by the following linear regression equation (1):
Plot of PMI versus vitreous potassium concentration Regression curve equation: y = 0.1733x + 2.3008 The two curves, drawn parallel to the regression line, represent 95% confidence interval for the regression line. PMI, postmortem interval
(y = potassium concentration in mmol/L, mean of the two eyes; x = PMI in hours)
The correlation coefficient (R = 0.962) meets the best correlations reported in the literature. Also, the equation parameters (slope and intercept) are substantially the same as those reported in a previous paper (y = 0.1698x + 2.3587) 17 calculated in an independent group of cases. This corroborates the reliability of the proposed method in both preanalytical and analytical aspects. The closeness of the correlation between vitreous potassium and PMI proves also the robustness of the method taking into consideration that samples were collected in different countries, stored for different times and in different locations, and shipped overseas.
The kinetics of postmortem potassium increase has also been studied by splitting the cases into two groups, one of which included PMIs ≤24 hours (group A, n = 105) and the other included PMIs >24 hours (group B, n = 59). The equation of the postmortem potassium increase calculated on group A was:
whereas the equation resulting from group B was:
The difference of the y intercepts can be considered simply due to an excessive extrapolation of the intercept on the y axis in group B data plot. On the other hand, the difference of slopes could be explained by hypothesizing a lower release of intracellular potassium in the early postmortem period (≤24 hours) from the still undamaged cell membranes.
The re-calculation of PMIs using equation (1) resolved for the variable x(PMI) produced data (calculated PMI) which differ from the observed PMIs by a mean error of 5.54 hours (SD = 4.16).
By plotting observed PMI versus the percent PMI calculation error ([observed PMI-calculated PMI]/observed PMI × 100), a drastic reduction of the percent calculation error with the increase of PMI is evident (Figure 3). In particular, in the first 24 hours the average percent calculation error is 41.62% (SD = 39.96), whereas between 24 and 109 hours the average error is 15.74% (SD = 12.09). This difference evaluated with Student's t-test is highly significant (P < 0.001). This observation is consistent with literature data.2,3
Plot of observed PMI versus percent PMI calculation error. PMI, postmortem interval
The reason for this phenomenon can easily be attributed to the interindividual variability of potassium concentrations in the vitreous humour at the time of death. This variability clearly affects the prediction of PMI in the early postmortem hours, when the overall potassium concentration in eye fluids is low, but later, when potassium concentration increases, it becomes less relevant.
Conclusion
On the basis of the data discussed above it can be concluded that the practical use of vitreous potassium for inferring the time of death is worthless in the first 24 hours, because of an unacceptable calculation error. However, in this range of PMI other parameters are available, particularly the decrease of body temperature and the fixation of hypostases.
On the other hand, after 24 hours, when the body temperature is equilibrated with that of the environment and hypostases are fixed, the potassium concentration increase in the eye fluid may become a valuable indicator of the time elapsed since death, being more accurate than any other physical findings, such as rigor mortis and the early putrefaction signs. According to our data, the usefulness of postmortem vitreous potassium lasts at least until 100 hours, after which the slope of the curve tends to a plateau and therefore loses information power (data not shown).
Last but not least, it must be stated that in the analysis of postmortem samples capillary electrophoresis may provide more accurate data than current clinical chemistry methods by providing analyte separation before quantification and thus avoiding possible interferences occurring in forensic samples and minimizing the calculation errors of PMI.
Footnotes
Acknowledgements
The present study was supported by ‘Fondazione Miria and Amleto Loro’ and by Governmental Funds PRIN - year 2007-2009 - # 2007XRNNRJ.