Applications of Tandem Mass Spectrometry (LC-MSMS) in estimating the post-mortem interval using the biochemistry of the vitreous humour

Applications of Tandem Mass Spectrometry (LC–MSMS) in estimating the post-mortem interval using the biochemistry of the vitreous humour Elena Lendoiroa, Cristina Cordeirob,c, María S. Rodrıíguez-Calvod, Duarte N. Vieirab,c, José M. Suárez-Peñarandad, Manuel López-Rivadullaa,d, José I. Muñoz-Barúsa,d,* a Institute of Forensic Science, University of Santiago de Compostela, Spain b National Institute of Legal Medicine and Forensic Sciences of Portugal – Centre Branch, Portugal c Centre of Forensic Sciences (CENCIFOR), University of Coimbra, Portugal d Department of Pathology and Forensic Science, University of Santiago de Compostela, Spain *Corresponding author: [email protected] ABSTRACT It is widely accepted that the quantification of potassium concentrations ([K+]) and hypoxanthine ([Hx]) in the vitreous humour is useful in estimating the time of death within a recent time interval. Despite all the advances made in this area, it is well recognized that difficulties in calibration, validation and the use of different methodologies and instrumentation from different laboratories may lead to obtaining different concentrations from the same sample. The extraction of the vitreous humour itself should also be carried out with a precise technique, taking care during the procedure to avoid causing any vascular injury that might lead to haematic contamination. Any fluid that is not clear could lead to erroneous results and should be discarded. We present a new LC–MSMS method developed for quantitative and qualitative analysis of [Hx] (valid also for guanine and xanthine), and qualitative determination of uric acid in vitreous humour. We also introduce a methodology to assess haematic contamination in order to improve the estimation of time since death. The method was fully validated in terms of linearity, sensitivity, imprecision, analytical recovery, extraction and process efficiency and matrix effect. Keywords: Tandem mass spectrometer; Post mortem interval; Hypoxanthine; Vitreous humour 1. Introduction In the field of forensic pathology one of the most important issues is the correct estimation of the post mortem interval (PMI). This is of particular relevance in criminal investigation, where recourse to a method that provides a PMI whose precision can be upheld in a court of law can be crucial [1,2]. It is widely accepted that the quantification of potassium concentrations ([K+]) and hypoxanthine ([Hx]) in the vitreous humour is useful in estimating the time of death within a recent time interval [3–7]. The importance of this problem is demonstrated by the fact that it has become the subject of numerous publications in leading journals in this field, which highlights the difficulties of this approach, and has led to further comparative studies [8]. It has been reported that urea (U) may modify the relationship between the concentration of vitreous potassium and PMI [4,9,10] and that the cause of death itself can also modify this relationship [10,11]. Different analytical methods for quantification of potassium and [Hx] have been developed in order to improve accuracy in estimating the time of death [12,13]. Although an inverse prediction by changing the variables is the statistical method of choice when using linear regression to relate the [K+] or [Hx] in the vitreous with PMI [4,14], other more flexible methods of regression have been developed which by including all the possible variables [K+], [Hx], [U], in the vitreous humour together with the cause of death, provide a more accurate estimate [11]. Although the calculation is much more complex than the simple application of the typical single-line regression formula, a free to use statistical package has been presented which provides a fast and intuitive result [15]. Despite all the advances made in this area, it is well recognized that the difficulties in calibration, validation, different methodologies and instrumentation from different laboratories may lead to obtaining different concentrations from the same sample [16,17]. It has also been demonstrated that pre-treatment of samples before analysis may lead to some variations in the determination of potassium and urea [17]. However, not only pre-treatment of the sample, but the extraction of the vitreous humour itself should be carried out with a precise technique, taking care during the procedure to avoid causing any vascular injury which might lead to haematic contamination. Any fluid that is not clear could produce erroneous results and should be discarded [2,4,9,18]. Liquid chromatography coupled to mass spectrometer (or tandem mass spectrometer) (LC–MS or LC–MSMS) has recently become increasingly important in post-mortem toxicology as a reliable technique for routine analysis. Unlike gas chromatography coupled to mass spectrometer (or tandem mass spectrometer) (GC–MS or GC–MSMS), it enables the analysis of polar compounds with no derivatization step, but has the same sensibility and specificity as GC–MSMS [19]. LC–MSMS has been employed for the identification and/or quantification of drugs of abuse, therapeutic drugs, poisons and/or their metabolites in post-mortem samples such as whole blood, plasma, serum, urine, cerebrospinal fluid, vitreous humour, liver and hair [20]. The aim of this present paper is to improve the estimation of PMI with vitreous biochemistry using a new analytical method, based on LC–MSMS, for the determination of Hx in vitreous humour (valid also for uric acid (UA), guanine (G), and xanthine (X)), and a methodology to assess haematic contamination in apparently clear samples of vitreous humour showing no haematic coloration due to the low presence erythrocytes in the sample. 2. Materials and methods 2.1. Chemical and reagents Hx, G, X and UA as a solid form were purchased by Acros Organics (Geel, Belgium). The internal standard (IS), 5-(p-methylphenyl)-5-phenylhidantoin, was supplied by EGA-Chemie (Steinheim, Germany). Acetonitrile and methanol were provided by Panreac Quimica S.A.U. (Barcelona, Spain). Ammonia solution (32%) and formic acid were supplied by Scharlau (Sentmenat, Spain). Ammonium acetate was supplied by Panreac (Barcelona, Spain). Purified water was obtained in the laboratory using a Milli-Q system (Le Mont-sur-Lausanne, Switzerland). OASIS MAX cartridges (60 mg, 3 mL) were from Waters (Mildford, USA). 2.2. Instrumentation The HPLC system consisted of a Waters Alliance 2795 Separation Module with a Waters Alliance series column heater/cooler (Waters Corp, Milford, USA). For the chromatographic separation, an Atlantis T3 (2.1 mm 100 mm, 3 mm) (Waters Corp, Milford, USA) analytical column was used at 30 8C. The mobile phase was ammonium acetate 10 mM (pH = 4.5) (A) and acetonitrile (B), working in gradient mode at a flow rate of 0.25 ml/min, and purge liquid was acetonitrile. The gradient applied was as follows: 0–2 min 100% A; 2–2.5 min from 100% to 20% A; 2.5–6.5 min 20% A; and 6.5–7 min from 20% to 100% A. The total run time was 14 min. For detection, a Quattro MicroTM API ESCI triple quadrupole (Waters Corp, Milford, USA) was employed. The instrument was operated in electrospray in positive mode (ESI+) to produce protonated molecules of the analytes under the following optimized settings: capillary voltage 3.0 kV; source block temperature 130 8C; desolvation gas (nitrogen) temperature 400 8C; desolvation gas flow rate 800 L/h; and cone gas (nitrogen) flow rate at 50 L/h. Data were acquired in MRM (multiple reaction monitoring) mode, which consists in the selection of the ion of interest (the precursor) in the first quadrupole (Q1), its fragmentation to form different product ions in the collision cell and finally the selection of a specific product ion in the third quadrupole (Q3). The association between the precursor and one product ion is referred as a transition. Transitions, cone voltage and collision energy were optimized by infusion of each individual analyte into the mass spectrometer (10 mM in NH4OH 2 mM) at 20 mL/min. Two precursor-product transitions per compound were monitored, except for IS, for which only one transition was selected. In Table 1, MRM transitions, cone voltage, collision energy and retention time (RT) are indicated for each analyte and IS. Data acquisition was controlled using MassLynx 4.0 software and processed with QuanLynx 4.0 software (Waters Corp, Milford, USA). 2.3. Calibrators and quality control preparation The primary standard solution of each compound was prepared separately after being weighed in solid form and reconstituted in ammonia solution (32%); initial concentrations of Hx, G and X were 10 mM, and 5 mM for UA. Each solution was prepared daily. Working standard solutions were prepared in water by dilution of primary solutions, and final concentrations were 1000, 200 and 100 mM for Hx and UA, and 100, 20 and 10 mM for G and X. The primary IS solution was prepared in methanol at 10 mM, and the working IS solution was prepared by dilution in methanol at 1 mM. Calibration curves (10–400 mM for Hx and UA, and 1–40 mM for G and X) were prepared by adding 15, 30 and 60 mL of the corresponding working solution, and 25 mL of working IS solution (1 mM). The analytes were endogenous compounds, consequently variable concentrations were found in control vitreous humour; for this reason calibration curves were prepared in 150 mL of water. Quality controls at low (10 mM for [Hx], 1 mM for G, 2 for X, and 20 for AU), medium (50 mM for Hx and UA, and 5 mM for G and X) and high (150 mM for Hx, 25 mM for G and X, and 250 mM for UA) concentrations were prepared daily in water from working solutions separate from those used for calibrators. 2.4. Sample preparation Vitreous humour samples were initially centrifuged for 10 min at 14,500 rpm in a Mini Spin Plus (Eppendorf, Hamburg, Germany). 2 mL of ammonium hydroxide 2 mM and 25 mL of IS were added to an aliquot of 150 mL of vitreous humour, previously centrifuged. Samples were submitted to a solid-phase extraction with OASIS MAX cartridges (Waters, Mildford, USA). After conditioning with 2 mL methanol and 2 mL water, samples were applied directly into the SPE column. Cleanup was accomplished by sequential washes with 2 mL 5% ammonium hydroxide in water and 2 mL 5% ammonium hydroxide in methanol. Cartridges were dried for 10 min under vacuum. Elution was performed in two sequential steps: first elution with 2 mL 2% formic acid in water, and second elution with 2.5 mL formic acid in methanol. Eluates were collected in the same tube and evaporated to dryness with nitrogen at 40 8C. Finally samples were reconstituted in 100 mL ammonium acetate 10 mM (pH = 4.5), and 30 mL were injected into LC– MSMS. 2.5. Validation The method was submitted to a full validation, including linearity, limit of detection (LOD), limit of quantification (LOQ), imprecision, analytical recovery, extraction efficiency, process efficiency, and matrix effect. Selectivity was not evaluated because it was not possible to obtain blank vitreous humour matrixes (i.e. without Hx, G, X and UA). Linearity was determined by the analysis of four calibration curves prepared on 4 different days. These calibration curves were prepared in water, and least-squares regression was applied. Acceptable linearity was achieved if coefficient of determination (r2) was at least 0.99, and calibrator residual was 20% at the LOQ and 15% at the other concentration levels. The LOD was defined as the lowest concentration with acceptable chromatography, and the presence of the two transitions with a signal-to-noise ratio of at least 3. The LOQ was the lowest concentration with a signal-to-noise of at least 10, imprecision lower than 20%, and analytical recovery between 80% and 120%. Imprecision and analytical recovery were determined at low, medium and high QC levels, with 5 replicates on 4 different days (n = 20), prepared in water. Imprecision, expressed as coefficient of variation (CV) of the measured values, was expected to be less than 15%. Krouwer and Rabinowitz guidelines [21] were followed for calculation of pooled intra-day, inter-day and total imprecision. The analytical recovery was evaluated as the percentage of the target concentration (n = 20), with an acceptance criterion of 85–115%. Extraction efficiency, process efficiency and matrix effect were determined using real specimens of vitreous humour at two levels of concentration (low and high) for Hx, G and X; and at one level of concentration for UA and IS. However, analytes are endogenous compounds, which are present in significant concentration in vitreous humour specimens. For this reason, specimen concentrations were previously measured (‘‘base value’’); and these ‘‘base values’’ were subtracted from the values of the same samples after the addition of 20 or 60 mM of Hx, 100 mM of UA, and 2 or 6 mM of G and X (‘‘spiked value’’). Extraction efficiency was calculated comparing the value obtained after subtraction of average peak areas of ‘‘base value’’ and ‘‘spiked value’’ of vitreous humour specimens fortified prior to extraction (n = 6) with those obtained after subtraction in specimens fortified after extraction (n = 6) at the same concentration. Process efficiency was determined by comparing the value obtained after subtraction of average peak areas of ‘‘base value’’ and ‘‘spiked value’’ of vitreous humour specimens fortified prior to extraction (n = 6), with peak areas of samples at the same nominal concentrations prepared in initial mobile phase (‘‘unextracted samples in mobile phase’’). Matrix effect was assessed by comparing analyte peak areas after subtraction of ‘‘base values’’ in 6 different vitreous humour extracted samples fortified after extraction with analyte peak areas of ‘‘unextracted samples in mobile phase’’. Good quantification of real samples with calibration curves prepared without matrix was tested by the analysis of five vitreous humours before and after adding 100 mM of Hx and UA, and 10 mM of G and X. The differences between these two measures were evaluated in terms of precision (calculated as %CV) and accuracy (expressed as inaccuracy and as the percentage of the mean relative error (%MRE)). 2.6. Evaluation effect of blood contamination Accidental puncture of the blood vessels in the vitreous humour during sample collection can alter [Hx] values, leading to an erroneous PMI interpretation. To test the effect of blood contamination, real specimens were contaminated using rising blood concentrations and the results compared with blood-free matrix from the same source. To simulate blood contamination, a blood dilution at 5% was prepared in saline solution, and 1, 2, 3, 4, and 5 mL of this dilution were added to different aliquots (150 mL) of the same vitreous humour. These results were compared with the values obtained from the same uncontaminated vitreous humour. This methodology was applied to 7 different vitreous humours, with 5 levels of blood contamination for each one. Finally, we performed a microscopic control of haematic contamination of the different aliquots by Pap staining after concentration by ultracentrifugation. 3. Results and discussion 3.1. Validation Linearity was verified for all the compounds except UA, where the results were not sufficiently satisfactory to determine its coefficient (r2) and residuals. The calibration curves of the rest of the compounds were fitted with 1/x2-weighted factor, and the r2 were above 0.99. The residuals were within 20% at LOQs and 15% at other calibrator concentrations for all compounds. The LOD ranged between 0.25 and 2.5 mM, and the LOQ was the lowest point of the calibration range. These results are summarized in Table 2. The LOD and LOQ values achieved with this method are appropriate for Hx determination, given that the values for this analyte found in previously papers ranged between 10.67 mM to 125.27 mM in 176 cases analyzed by Muñoz et al. [9] and between 4.8 mM and 45.6 mM in 120 cases analyzed by Jashnani et al. [7]. Table 3 shows the results of imprecision and analytical recovery. These were satisfactory for all tested compounds except UA, which had problems with linearity. Table 4 shows extraction efficiency, process efficiency and matrix effect results of Hx, G, X, UA and IS. Extraction efficiency ranged from 62.6% to 114.4%, with values higher than 50% for all analytes. Hx, G and IS showed ion enhancement ranging from 14.1% to 137.1%, and UA ion suppression. X showed no significant matrix effect. In all cases % CV was below 20%. Process efficiency ranged from 25.3% from 127.6%. Good quantification of real specimens with non-matrix calibration curves was achieved for Hx, G and X. Adequate accuracy was found, with values of %MRE below 15% (-0.3% for Hx, -12.8 for G and -3.3 for X). Precision results (%CV was 7.7 for Hx, 12.7 for G and 17.4 for X) were satisfactory for Hx and G, with values below 15%, and slightly higher for X, but below 20%. Very few previous papers report [Hx] determination by LC– MSMS [22,23], and these cases used urine samples. No previous methods have been developed for determination of this analyte in vitreous humour, despite the high sensitivity and specificity of this technique. 3.2. Evaluation effect of blood contamination Macroscopically, the samples showed no coloration although the vitreous humours were contaminated with blood. Microscopically, after staining the aliquots, there was no evidence of erythrocytes. These samples seem to show a decrease in Hx concentrations, as well as in G and X concentrations, in relation to increased blood concentrations. However, no significant correlation between the 5 levels of blood contamination was observed in any of the 15 samples analyzed. Table 5 shows the values obtained for the selected analyzed samples of vitreous humour and their level of blood contamination. Applying the formula proposed [12] for determining PMI, and given the value of [Hx], the change in the central value of PMI of the confidence interval estimated can cause a variation in a PMI estimate from 14.4 h for an uncontaminated sample to 8.6 h for a contaminated one (see Table 5, Specimen 1. The mean variation for total specimens was 4.02 h). No statistically significant relationship between the increase in blood concentration and changes in the estimate were found, which further demonstrates the impossibility of estimating PMI by means of determining blood Hx because an erroneous PMI value would be obtained from contaminated samples. Such modifications also affect G and X (Table 5). However, UA showed characteristic behaviour and all samples with blood contamination showed a total disappearance of the two MRM transitions of this analyte. Although UA validation was not satisfactory, in terms of linearity, analytical recovery and precision, it was possible to make a qualitative determination. Fig. 1 presents the chromatogram of a real specimen of vitreous humour before and after blood contamination with 1 mL of blood dilution in saline solution at 5%. A total disappearance of UA was clearly observed. This method makes it possible to identify those samples where blood contamination is due to improper extraction of the vitreous humour. Even at low levels of contamination (when not microscopically evident), and before Hx concentrations are affected, the two MRM transitions of UA disappear, thus enabling the samples to be discarded, in the same manner as the clearly contaminated samples, and thus avoid erroneous results in PMI [2,4,9,18]. The procedure has been applied in the routine analysis of hypoxanthine in vitreous humour. A total of 186 clear vitreous, obtained by well-trained forensic pathologist with long experience in obtaining the samples, were analyzed. Of these, 7 showed total disappearance of the two transitions of uric acid although no macroscopically blood contamination was observed, proving the relevance of using this new analytical procedure. 4. Conclusions A new LC–MSMS method was developed for quantitative and qualitative analysis of Hx, G and X, and qualitative determination of UA. The method was fully validated in terms of linearity, sensitivity, imprecision, analytical recovery, extraction and process efficiency, and matrix effect. During the process of vitreous humour extraction blood contamination can occur as a result of vascular injury and such contamination could distort the results of PMI. However, by monitoring UA it has been possible to establish a new criterion to identify specimens of vitreous humour with blood contamination, even when no coloration is evident. For these reasons it is a method of choice for the estimation of the PMI by means of the determination of Hx and where haematic contamination can be objectively demonstrated. Acknowledgments We gratefully acknowledge the support by INCITE (Consellería de Innovación e Industria, Xunta de Galicia), Project Number INCITE09 228 166 PR. Presented in part at the 19th Meeting of the International Association of Forensic Sciences, Funchal. References [1] W.Q. Sturner, G.E. Gantner Jr., The post-mortem interval. A study of potassium in the vitreous humor, Am. J. Clin. Pathol. 42 (1964) 137–144. [2] J.I. Coe, Vitreous potassium as a measure of the post-mortem interval: an historical review and critical evaluation, Forensic Sci. Int. 42 (1989) 201–213. [3] T.O. Rognum, S. Hauge, S. Oyasaeter, O.D. Saugstad, A new biochemical method for estimation of post-mortem time, Forensic Sci. Int. 51 (1991) 139–146. [4] J.I. Muñoz, J.M. Suárez-Peñaranda, X.L. Otero, M.S. Rodríguez-Calvo, E. Costas, X. Miguéns, L. Concheiro, A new perspective in the estimation of post-mortem interval (PMI) based on vitreous, J. Forensic Sci. 46 (2001) 209–214. [5] C. Henssge, B. Knight, T. Krompecher, B. Madea, L. Nokses, Estimation of the time since death, in: The Early Post-mortem Period, 2nd edn, Arnold, London, 2002. [6] B. Madea, C. Henssge, Time since death, in: J. Payne-James, A. Busuttil (Eds.), Forensic Medicine: Clinical and Pathological Aspects, Greenwich Medical Media Limited, London, 2003, pp. 91–114. [7] K.D. Jashnani, S.A. Kale, A.B. Rupani, Vitreous humor: biochemical constituents in estimation of post-mortem interval, J. Forensic Sci. 55 (2010) 1523–1527. [8] N. Lange, S. Swearer, W.Q. Sturner, Human post-mortem interval estimation from vitreous potassium: an analysis of original data from six different studies, Forensic Sci. Int. 66 (1994) 159–174. [9] J.I. Muñoz, J.M. Suárez-Peñaranda, X.L. Otero, M.S. Rodríguez-Calvo, E. Costas, X. Miguéns, L. Concheiro, Improved estimation of post-mortem interval based on differential behaviour of vitreous potassium and hypoxanthine in death by hanging, Forensic Sci. Int. 125 (2002) 67–74. [10] B. Madea, C. Henssge, Eye changes after death, in: B. Knight (Ed.), The Estimation Time Since Death in the Early Post-mortem Period, 3rd edn, Edward Arnold, London, 1996, pp. 106– 137. [11] J.I. Muñoz-Barú s, M. Febrero-Bande, C. Cadarso-Suárez, Flexible regression models for estimating post-mortem interval (PMI) in forensic medicine, Stat. Med. 27 (2008) 5026–5038. [12] J.I. Muñoz, E. Costas, M.S. Rodríguez-Calvo, J.M. Suárez-Peñaranda, M. LopezRivadulla, L. Concheiro, A high performance liquid chromatography method for hypoxanthine determination in vitreous humour: application to estimation of post mortem interval, Hum. Exp. Toxicol. 25 (2006) 279–281. [13] M.L. Passos, A.M. Santos, A.I. Pereira, J.R. Santos, A.J. Santos, M.L. Saraiva, J.L. Lima, Estimation of post-mortem interval by hypoxanthine and potassium evaluation in vitreous humor with a sequential injection system, Talanta 79 (2009) 1094–1099. [14] B. Madea, A. Rödig, Time of death dependent criteria in vitreous humor: accuracy of estimating the time since death, Forensic Sci. Int. 164 (2006) 87–92. [15] J.I. Muñoz-Barús, M.S. Rodríguez-Calvo, J.M. Suárez-Peñaranda, D.N. Vieira, C. Cadarso-Suárez, M. Febrero-Bande, PMICALC: an R code-based software for estimating post-