Jorgensen_jatnd10.qxd:JATLynneTemplate

10/19/10

3:16 PM

Page 1

Journal of Analytical Toxicology, Vol. 34, November/December 2010

Determination of Ketone Bodies in Blood by Headspace Gas Chromatography–Mass Spectrometry Karen Marie Dollerup Holm*, Kristian Linnet, Brian Schou Rasmussen, and Anders Just Pedersen Section of Forensic Chemistry, Department of Forensic Medicine, Faculty of Health Science, University of Copenhagen, Frederik V’s vej 11, DK-2100, Copenhagen, Denmark

Abstract A gas chromatography–mass spectrometry (GC–MS) method for determination of ketone bodies (ββ-hydroxybutyrate, acetone, and acetoacetate) in blood is presented. The method is based on enzymatic oxidation of D-ββ-hydroxybutyrate to acetoacetate, followed by decarboxylation to acetone, which was quantified by the use of headspace GC–MS using acetone-13C3 as an internal standard. The developed method was found to have intra- and total interday relative standard deviations < 10% for acetone+acetoacetate levels (~25 to 8300 µM) and D-ββhydroxybutyrate levels (~30 to 16500 µM). Recovery values varied from 98 to 107%, demonstrating the suitability of the method for measuring ketone bodies over a wide concentration range. The method has been applied to cases in which ketoacidosis was suspected as the cause of death in diabetics or chronic alcoholics, as well as to cases in which another cause of death was identified.

Introduction Forensic medicine is frequently challenged by cases of sudden death in chronic alcoholics. At times, the cause of death cannot be determined by autopsy, histological examination, alcohol, and/or drug testing. Alcoholic ketoacidosis, a consequence of alcohol-induced hypoglycemia typically during abstinence periods with low food intake, must be considered in such cases (1). During ketoacidosis, a large increase of the ketone bodies acetone (Ac), acetoacetate (AcAc), and β-hydroxybutyrate (βHB) occurs, with the latter two molecules affecting the blood bicarbonate concentration and pH (2). Ketoacidosis can result from diabetic ketoacidosis (type I diabetics only), hypothermia, or starvation or represent alcoholic ketoacidosis, meaning that the case circumstances, autopsy, and toxicological findings should always be considered (2–5).

Measuring one or all of the ketone bodies is used as an indication of ketoacidosis, and in most cases an enzymatic oxidation of β-HB is performed based on the work done by Williamson et al. (6), Siegel et al. (7), and/or Felby et al. (8). The reaction scheme for the determination of Ac, AcAc, and βHB is shown in Figure 1. Following the oxidation of β-HB, the resulting AcAc is then decarboxylated to Ac, which can easily be measured. By examining the sample with and without the addition of the enzymatic reactants, the combined concentration of Ac and AcAc can be subtracted from the total ketone concentration, resulting in the β-HB concentration, which is the most abundant of the three ketone bodies (1,3,5,9). Two methods using derivatization for the determination of β-HB with gas chromatography–mass spectrometry (GC–MS) were recently published (10,11). Most previously published methods used GC or highperformance liquid chromatography (HPLC) in combination with ultraviolet (UV) or flame-ionization (FID) detection (8,12– 14). In this report, we describe an updated, modified method based on an earlier enzyme-based headspace (HS)-GC–FID method for measuring the ketone bodies in postmortem blood published by Felby et al. (8). To our knowledge, this is the first reported method using GC–MS in combination with enzymebased analysis of ketone bodies. We performed the decarboxylation of AcAc in the HS autosampler oven, ensuring reproducibility of the necessary temperature and time. Additionally, the current method employs an isotopic internal standard (IS), an improvement over previous IS use because of the almost identical properties to the analyte (7,12).

Materials and Methods Chemicals and reagents

* Author to whom correspondence should be addressed. Section of Forensic Chemistry, Department of Forensic Medicine, Faculty of Health Science, University of Copenhagen, Frederik V’s vej 11, DK-2100, Denmark. Email: [email protected].

Nicotinamide adenine dinucleotide (NAD) free acid (grade II, 98% purity), 3-hydroxybutyrate dehydrogenase type II from Rhodobacter spaeroides, and rabbit muscle lactate dehydrogenase were obtained from Roche (Mannheim, Germany). Di -

Reproduction (photocopying) of editorial content of this journal is prohibited without publisher’s permission.

549

Jorgensen_jatnd10.qxd:JATLynneTemplate

10/19/10

3:16 PM

Page 2

Journal of Analytical Toxicology, Vol. 34, November/December 2010

sodium hydrogen phosphate was of analytical grade and purchased from Merck (Darmstadt, Germany) and pyruvate, monosodium salt, crystalline was purchased from Fluka (Steinheim, Germany).

5975C inert XL MSD, all from Agilent Technologies (Santa Clara, CA). The capillary column was a DB-624 (30 m × 0.250mm i.d., 1.0-μm stationary phase, J&W, Agilent Technologies). HC-GC–MS analysis

Controls and standards

Acetone (Ac, > 99.8%, used as a standard) was purchased from Merck; Ac-13C3 (99% 13C, internal standard), Ac (> 99.5%, control), and racemic sodium 3-hydroxybutyric acid (> 99%, control) were purchased from Fluka; and racemic sodium (DL)-β-hydroxybutyric acid (~98%, standard) was obtained from Sigma (Steinheim, Germany). Working standards of 15000 µM Ac in Milli-Q water and 100 µM Ac-13C3 (IS) were prepared. Milli-Q water and the working standard were used for a one-point calibration. Control blood samples was prepared from blood obtained from the local blood bank, where low Ac levels (< 50 µM) were measured before use. The low control was spiked to levels of approximately 150 µM Ac and 250 µM β-HB, including the measured endogenous levels, and Ac and β-HB were added to the high control in concentrations of 16000 and 8000 µM, respectively. All working solutions were found to be stable for 6 months at 4°C (data not shown). Once a year, the current method is applied to two serum samples containing β-HB as part of an external quality assessment scheme (LabQuality, Helsinki, Finland).

We used the following HS parameters: oven temperature, 90°C; loop temperature, 110°C; transferline temperature, 115°C; pressurization pressure, 27 psi; pressurization time, 0.4 min; loop fill time, 0.2 min; loop equilibration time, 0.1 min; injection time, 1.0 min. Parameters used on the GC were column temperature, 50°C for 5 min followed by 70°C/min to 200°C, held for 2 min; inlet temperature, 250°C; carrier gas (He) constant pressure, 14 psi; split, 1:15. The interface between the GC and the MS is equipped with a quick-swap piece. The quick-swap pressure was 2 psi. The MS interface parameters were 250°C; quadrupole temperature, 150°C; MS source temperature, 230°C; and element energy, 70 V. Ions used for selected ion monitoring (SIM) were m/z 43 and 58 for Ac, and m/z 61 and 45 for the IS, Ac-13C3. A small fraction of the time was used for SCAN of m/z 20–200, to help indentify other compounds present. Data analysis was conducted by the Enhanced ChemStation, MSD Chemstation E.01.01.335 from Agilent Technologies, using the ions m/z 61 and 43 for quantification.

Results

Sample preparation

The sample preparation was conducted in two parts. For the measurements for Ac and AcAc, 100 µL blood or standard was added to 200 µL phosphate buffer solution (0.032 M, 8.5 pH), followed by 100 µL IS work standard in a 20-mL headspace vial. The vial was sealed with a PTFE-faced septum. The specimens were analyzed for Ac by HS-GC–MS. For determination of the total ketone body sum, 100 µL blood or standard was added to a 20-mL headspace vial, followed by 100 µL IS solution and 200 µL freshly mixed solution of phosphate buffer solution containing 0.3125 mg/mL D-βhydroxybutyrate dehydrogenase and lactate dehydrogenase, 0.05 M pyruvate, and 3.75 mM NAD. The vial was sealed with a PTFE-faced septum. After incubation at 37°C for 30–40 min, the specimens were analyzed for Ac by HS-GC–MS. Femoral blood taken during autopsy was stabilized with 100 mg sodium fluoride in a 10-mL container and used as a specimen. Serum, urine, and vitreous humor can also be analyzed by the method, treated in the same way as described for blood. The concentration of ketone bodies varies depending on the sample type (15,16).

Method optimization

The addition of an IS to the ketone body determination method eliminates error introduced by variations in injection and detection, improving the stability of the method. By choosing an IS with very similar properties to the analyte, the matrix effect will be eliminated, as the matrix effect will influence the analyte and the IS in an identical way. It was expected that Ac-13C3 was a well-suited IS, as it is more similar to Ac than Ac-d6 or other polar compounds, has an almost identical mass and Rt, and is not influenced by the exchange of hydrogen. Addition of salt, which was previously used (8), was found to have no effect on the calibration curve achieved with

Apparatus

The HS-GC–MS apparatus was a G1888 network headspace sampler coupled to an 7890A GC System with the

550

Figure 1. The reaction scheme used in this paper in the determination of the ketone bodies acetone, acetoatectate, and β-hydroxybutyrate.

Jorgensen_jatnd10.qxd:JATLynneTemplate

10/19/10

3:17 PM

Page 3

Journal of Analytical Toxicology, Vol. 34, November/December 2010

the IS, and the response of acetone in blood and water was found to be identical as well (data not shown). Detection with an MS detector provides better selectivity and sensitivity for the method, since compound-specific ions can be used for identification. Linearity was ensured by selecting the second most abundant ion for Ac-13C3 (m/z 61), and the most abundant ion for Ac (m/z 43), as quantification ions. No influence was seen on m/z 61 from the analyte, and the ion response was satisfactory, even at low IS concentration by the use of single ion monitoring (SIM). The ions m/z 45 and 58 were used as qualifier ions for the IS and analyte, respectively. In order to determine the necessary HS parameters for full recovery of AcAc, the time dependence of the decarboxylation of AcAc to Ac in blood (10000 µM) was examined at HS oven

temperatures of 100, 90, and 80°C (Figure 2). Our observations agree with those presented previously (8), showing that the decarboxylation was complete after 60 min at 100°C; 90°C also gave a satisfactory result, but with a lower peak area as a consequence of the lower temperature. To increase the range of the method, the capacity of the enzymatic reaction was examined. Racemic β-HB was used for standard and control preparation, but only D-β-HB is oxidated by the enzymatic reaction (data not shown). The blood sample was here diluted 1:3 during the sample preparation, instead of the previously used 1:1 ratio (8), before examinations were conducted, to increase the pyruvate/β-HB ratio. The time dependence and capacity of the enzymatic oxidation of β-HB were examined using concentrations of 4000 and 20000 µL βHB; 30 min was sufficient for full recovery at both concentrations (Figure 3). The full oxidation of β-HB and decarboxylation of AcAc mean that quantification of all three compounds can be performed based on an Ac calibration curve. This conclusion was verified by analysis of Ac- and β-HB-based calibration curves, which were found to be identical to eachother. In the final method, the use of controls containing β-HB ensured full oxidation and decarboxylation in the examined samples. Method performance

Figure 2. Time dependence of the decarboxylation of 10000 μM acetoacetate at 100, 90, and 80°C.

Figure 3. Time dependence of the enzymatic oxidation of β-hydroxybutyrate at 37°C, replicate measurements at 0 and 90 min at each concentration of β-hydroxybutyrate, four measurements at 30 and 60 min at each concentration of β-hydroxybutyrate.

The method was validated according to Peters et al. (17) and in-house procedures. The validation parameters considered were limit of detection (LOD), limit of quantification (LOQ) selectivity, carryover, sample stability, linearity, accuracy, and precision. The selectivity was examined in control samples spiked with common volatile compounds (e.g., ethanol, isopropanol, methanol, and acetaldehyde), and also on authentic samples, revealing chromatographic separation and no ion interference. Full separation of Ac from common blood volatiles was obtained by GC. The chosen IS and Ac elute at the same time as a result of their similarity in mass and physiochemical properties, but the use of MS and selected quantification and qualification ions for the two compounds resulted in good selectivity. Chromatograms from an authentic sample with a large amount of volatiles are shown in Figure 4. The linear Ac range was examined twice by an eight-point calibration curve in triplicate over the range 10–25000 µM. Linearity was confirmed over 10–25000 µM, with a correlation coefficient of 0.999

551

Jorgensen_jatnd10.qxd:JATLynneTemplate

10/19/10

3:17 PM

Page 4

Journal of Analytical Toxicology, Vol. 34, November/December 2010

with residuals < 10% over the full range, using 1/x weighing of the data in a first-degree polynomial (data not shown). A small positive off-set was observed as a result of small Ac impurities in the IS and water. The standard deviation (SD) determined at levels of ~ 30 µM and ~ 55 µM (n = 16) was used to calculate the limits of deTable I. Intraday Precision (n = 4) and Total Interday Precision (n = 2) and Accuracy (n = 8) Analyzing Ketone Bodies by HS-GC–MS Method Level µM

Intraday %RSD

Interday %RSD

Accuracy %

Acetone + acetoacetate ~30 ~290 ~1100 ~5300 ~8300

1.4 1.7 0.4 1.4 1.3

4.7 4.0 6.1 4.3 3.1

– 107 106 104 105

β-Hydroxybutyrate ~25 ~540 ~2100 ~10500 ~16500

3.4 1.7 2.8 1.2 1.2

3.4 3.8 3.5 4.8 3.6

– 100 98 101 100

tection (LOD) (3 × SD + blank) and quantification (LOQ) (10 × SD + blank), finding them to be 8 and 21 µM in blood, respectively. The accuracy and intra- and total interday precisions were determined by measuring spiked blood at five levels of β-HB (~25 to 16500 µM) and Ac (~30 to 8300 µM) four times by two different operators on two identical HS-GC–MS systems on two different days. The intraday and total interday precisions were calculated as the relative standard deviation in percentage (%RSD) using one-way ANOVA. The accuracy of the spiked samples was calculated as the determined mean concentration against the theoretically spiked concentration plus the determined physiological values in the used blood. Table I demonstrates that the intra- and interday precisions at all levels were less than 15%. The interday %RSD varied from 3.4 to 4.8 for β-HB, and 3.1 to 6.1 for Ac+AcAc and can be explained by the intraday variation. Internal quality controls (QC) at a high and low level were run over a period of 8–9 days and also resulted in %RSD < 10 (Table II), well below the recommended %RSD < 15(17). The accuracy varied from 104% to 107% and 98% to 101% for Ac+AcAc and β-HB, respectively, which are all within the recommended 100% ± 15%(17). Water samples with 25000 µM Ac were used to determine a carryover below 0.03%. The stability of 10 authentic samples prepared for analysis was examined after storage at room temperature for 24 h (in the autosampler); only minor changes were found in determined Ac+AcAc level (< 10% or < 29 µM increase) and β-HB (< 10% or < 22 µM loss). The total ketone body concentration was only slightly affected, with changes < 7% or < 17 µM. It has been found that prolonged heating of the samples in the HS oven (90°C) gives unreliable results (data not shown). The method was applied to two serum samples containing β-HB in an external quality assessment scheme, and was within one SD of the overall mean (Table III). Method application

Figure 4. Chromatograms achieved with the used method applied to an authentic case without enzyme and IS addition: scan chromatogram for m/z 20–200 (A) and quantification and qualifying ion for acetone, m/z 43 and m/z 58 (B). Relevant peaks are labeled.

552

The method was applied to 21 authentic cases, 14 in which the subject had died from physical injuries, with 5 of these subjects known alcoholics and/or diabetics. Four cases of suspected alcoholic ketoacidosis and three diabetic ketoacidosis were examined. A case was identified as suspected alcoholic ketoacidosis when no ascertainable cause of death was found in a known alcoholic with a fatty liver. Cases where high glucose levels were found in known diabetics with no other ascertainable cause of death were categorized as cases of suspected diabetic ketoacidosis. The results are given in Table IV.

Jorgensen_jatnd10.qxd:JATLynneTemplate

10/19/10

3:17 PM

Page 5

Journal of Analytical Toxicology, Vol. 34, November/December 2010

Discussion A reliable and easy method for the determination of ketone bodies is needed, especially in suspected cases of diabetic or alcoholic ketoacidosis. We present a method based on the method of Felby et al. (8), which has been used for years by our department and by several others (3,12,18). The Felby et al. (8) method has proven to be reliable, but given its limited range for β-HB (1–700 µM), reanalysis of samples was a frequent issue. By increasing the range up to 16000 µM β-HB (24000 µM total) from the original 700 µM, the number of reanalyzed cases has been reduced from 22% to < 2% (based on all cases analyzed at the department in 2008, n > 260). The increase in range is partially connected to the decrease in sample size, as the enzymatic reaction will produce full recovery at higher blood concentrations. This decreased sample amount is also an advantage of our updated method. Manual integration was also performed regularly, as full chromatographic separation of Ac was not always achieved by the previously used method. An IS was not used, and NaCl was added to the samples in the original method to eliminate potential errors caused by variation in blood samples (decomposition, etc.). By moving the method from HS-GC–FID to HS-GC–MS, and by adding an IS, this method has been modernized and now demonstrates increased selectivity. For the application of the method in cases where ketoacidosis is expected, the LOQ of 21 µM is satisfactory, as lethal Table II. Precision of Analysis of In-House Quality Control Samples with Low Concentrations (n = 9) and High Concentration QC (n = 8) of Ketone Bodies with All Measurements Made in Duplicate Level µM

Interday %RSD

Acetone + Acetoacetate ~150 ~5100

9.5 5.8

β-Hydroxybutyrate ~240 ~10300

4.5 4.4

levels are set to be several orders above varying from 531 µM of the total concentration to 3000 µM β-HB, depending on the publication (14,15,18,19). The low LOQ makes it possible to use the new method in clinical studies, in which accurate measurements of control groups will be necessary. The interday %RSD of the new method varies from 3.1% to 6.1%, a slight improvement over the original method and similar to other methods (8,20). This new method gives a better overall accuracy than the original Felby et al. (8) method (83– 100%), even though a single Ac-based one-point calibration curve is applied. In particular, the accuracy for β-HB (98– 101%) is much higher than that resulting from the Felby et al. (8) method (83–91%). This increase in accuracy is likely a result of complete oxidation and decarboxylation of β-HB, which was not achieved by Felby et al. (8). The accuracy varies from 98% to 101% for β-HB, supporting our method development conclusion that enzymatic oxidation and decarboxylation go to completion under the given conditions and the use of β-HB in the controls offers an indirect control of AcAc determination. The controls showed similar values to those found in the method validation. The accuracy is verified by the external quality assessment data, where the measured values are within the one SD of the mean. The change from FID to MS detection has improved specificity, as the ion ratio can be used as an identification paramTable III. Results from External Quality Assessment Scheme for Determination of β-Hydroxybutyrate in Serum Provided by LabQuality (μM) Method

n

Mean

SD

Sample 1 Overall Abbott bedside tests Enzymatic tests Presented HS-GC–MS

14 7 6 1

1180 1330 1030 1050

180 50 110

Sample 2 Overall Abbott bedside tests Enzymatic tests Presented HS-GC–MS

15 8 6 1

4050 4410 3620 3800

520 400 300 –

Table IV. Ketone Body Concentrations (µM) Measured in Postmortem Blood from Four Groups of Subjects Showing Average and Range of Values Group

n

Ac+AcAc (µM)

β-HB (µM)

Total (µM)

1: Control

9

66 (41–130)

84 (43–260)

149 (91–336)

2: Alcoholics/diabetic

5

85 (57–120)

271 (18–730)

356 (89–850)

3: Alcoholic ketoacidosis

4

1630 (1100–2200)

4074 (1540–6000)

5704 (2840–8200)

4: Diabetic ketoacidosis

3

5270 (1600–8400)

16333 (14000–20000)

24000 (15600–28400)

553

Jorgensen_jatnd10.qxd:JATLynneTemplate

10/19/10

3:17 PM

Page 6

Journal of Analytical Toxicology, Vol. 34, November/December 2010

eter in addition to the retention time. SIM allows the use of an isotopic IS, which helps ensure an improvement in accuracy. The need for salt addition is made unnecessary, as the matrix effect is eliminated by the presence of the IS, making the sample preparation simpler than the original method. The prepared sample changes are not significant enough to effect the interpretation of the ketone bodies, and samples left in the HS autosampler overnight can be used. Based on the examination of suspected ketoacidosis samples, and cases resulting from other causes of death, our method is suitable for use in forensic work. The observed mean values correspond well with values in previous reports (1,14,15,19– 21). The higher concentrations of ketone bodies found in people with a history of diabetes or alcoholism compared to the control group fit observations made in other studies (15,18,20,22). Even though only three cases of diabetic ketoacidosis are included in the present study, the observation that the ketone body concentrations are generally higher than in the case of alcoholic ketoacidosis agrees with a prior study (14).

Conclusions We have developed an accurate and precise method for the determination of ketone bodies in postmortem blood. This method is the first published enzyme-facilitated HS-GC–MS method for ketone bodies, and it has been found suitable for application in forensic casework when ketoacidosis is suspected.

References 1. L.N. Denmark. The investigation of beta-hydroxybutyrate as a marker for sudden-death due to hypoglycemia in alcoholics. Forensic Sci. Int. 62: 225–232 (1993). 2. O.E. Owen, S. Caprio, G.A. Reichard, Jr., M.A. Mozzoli, G. Boden, and R.S. Owen. Ketosis of starvation: a revisit and new perspectives. Clin. Endocrinol. Metab. 12: 359–379 (1983). 3. G. Teresinski, G. Buszewicz, and R. Madro. The influence of ethanol on the level of ketone bodies in hypothermia. Forensic Sci. Int. 127: 88–96 (2002). 4. J. Mesa, D. Salcedo, H. de la Calle, E. Delgado, J. Novoa, F. Hawkins, G.S. Navarrete, M. Parramon, and D. Acosta. Detection of ketonemia and its relationship with hyperglycemia in type 1 diabetic patients. Diabetes Res. Clin. Pract. 72: 292–297 (2006). 5. B. Brinkmann, G. Fechner, B. Karger, and A. DuChesne. Ketoacidosis and lactic acidosis—frequent causes of death in chronic alcoholics? Int. J. Legal Med. 111: 115–119 (1998).

554

6. D.H. Williamson, H.A. Krebs, and J. Mellanby. Enzymic determination of D-(–)-beta-hydroxybutyric acid and acetoacetic acid in blood. Biochem. J. 82: 90–96 (1962). 7. L. Siegel, N.I. Robin, and L.J. Mcdonald. New approach to determination of total ketone bodies in serum. Clin. Chem. 23: 46– 49 (1977). 8. S. Felby and E. Nielsen. Determination of ketone-bodies in postmortem blood by headspace gas-chromatography. Forensic Sci. Int. 64: 83–88 (1994). 9. J.R. Steinhauer, A. Volk, R. Hardy, R. Konrad, T. Daly, and C.A. Robinson. Detection of ketosis in vitreous at autopsy after embalming. J. Forensic. Sci. 47: 221–223 (2002). 10. H.M. Hassan and G.A. Cooper. Determination of beta-hydroxybutyrate in blood and urine using gas chromatography–mass spectrometry. J. Anal. Toxicol. 33: 502–507 (2009). 11. S. Elliott, C. Smith, and D. Cassidy. The post-mortem relationship between beta-hydroxybutyrate (BHB), acetone and ethanol in ketoacidosis. Forensic. Sci. Int. 198(1-3): 53–57 (2010). 12. D.J. Pounder, R.J. Stevenson, and K.K. Taylor. Alcoholic ketoacidosis at autopsy. J. Forensic Sci. 43: 812–816 (1998). 13. Y.S. Hsu and S.H. Chen. Trace analysis of beta-hydroxybutyrate in human plasma by derivatization and high-performance liquid chromatography. Chromatographia 57: 735–739 (2003). 14. P.X. Iten and M. Meier. Beta-hydroxybutyric acid—an indicator for an alcoholic ketoacidosis as cause of death in deceased alcohol abusers. J. Forensic Sci. 45: 624–632 (2000). 15. P. Kadis, J. Balazic, and V. Ferlan-Marolt. Alcoholic ketoacidiosis: a cause of sudden death of chronic alcoholics. Forensic Sci. Int. 103: S53–S59 (1999). 16. S. Felby, E. Nielsen, and J.L. Thomsen. The postmortem distribution of ketone bodies between blood, vitreous humor, spinal fluid, and urine. Forensic Sci. Med. Pathol. 4: 100–107 (2008). 17. F.T. Peters, O.H. Drummer, and F. Musshoff. Validation of new methods. Forensic Sci. Int. 165: 216–224 (2007). 18. J.L. Thomsen, S. Felby, P. Theilade, and E. Nielsen. Alcoholic ketoacidosis as a cause of death in forensic cases. Forensic Sci. Int. 75: 163–171 (1995). 19. J. Kanetake, Y. Kanawaku, S. Mimasaka, J. Sakai, M. Hashiyada, M. Nata, and M. Funayama. The relationship of a high level of serum beta-hydroxybutyrate to cause of death. Leg. Med. 7: 169– 174 (2005). 20. S. Yamato, K. Shinohara, S. Nakagawa, A. Kubota, K. Inamura, G. Watanabe, S. Hirayama, T. Miida, and S. Ohta. High-performance liquid chromatography determination of ketone bodies in human plasma by precolumn derivatization with p-nitrobenzene diazonium fluoroborate. Anal. Biochem. 384: 145–150 (2009). 21. E. Osuna, G. Vivero, J. Conejero, J.M. Abenza, P. Martinez, A. Luna, and M.D. Perez-Carceles. Postmortem vitreous humor beta-hydroxybutyrate: its utility for the postmortem interpretation of diabetes mellitus. Forensic Sci. Int. 153: 189–195 (2005). 22. D.J. Pounder. Problems in the necropsy diagnosis of alcoholic liver-disease. Am. J. Forensic Med. Pathol. 5: 103–109 (1984).

Manuscript received February 24, 2010; revision received May 4, 2010.