Evaluation of the Methods Used for Carboxyhemoglobin Analysis in Postmortem Blood

Abstract

Numerous methods have been described in the literature for the determination of carboxyhemoglobin (COHb) in whole blood. The most popular and widely used have been (1) the spectrophotometric methods, which could be performed either by using a conventional spectrophotometer or by using specialized automated instruments known as CO-oximeters; (2) the gas chromatographic methods, with variable detection systems, which have been considered as the reference methods for every carbon monoxide analysis. The authors have critically reviewed previously reported comparative studies on these methods, considering statistical and analytical matters, in order to propose the best method for the determination of COHb in postmortem blood, that could be utilized in forensic toxicology laboratories where such analyses are limited in number (less than 20 per year). Criteria for evaluation have been accuracy, reliability, simplicity, time, and cost. The authors’ concluding statement has been that the manual spectrophotometric method could be the method of choice for COHb determination in postmortem blood samples. It is simple, rapid, and reliable and fulfills the forensically acceptable accuracy. It is performed by the use of a conventional spectrophotometer, which is considered a basic instrument in every analytical laboratory.

Carbon monoxide (CO) has been considered responsible for a large percentage of the accidental poisonings and deaths reported throughout the world each year (Raub et al. 2000). Fatal cases could have been grossly underreported or misdiagnosed by medical professionals (Raub et al. 2000; Heckerling et al. 1990). As CO poisoning has been a diagnosis frequently overlooked, the importance of measuring carboxyhemoglobin (COHb) in suspicious settings cannot be overstated (Raub et al. 2000; Koumbourlis and Skoutakis 1982). Furthermore, as CO has been considered a major factor in fire deaths, aircraft accidents, and intentional exposure to autoexhaust, COHb has been a commonly determined substance in forensic toxicology laboratories, in order to assist the medical examination, or in the elucidation of the cause and manner of death.

In forensic cases the interpretation of COHb saturation levels could be defined as follows: values above 50% COHb have been usually consistent with death. When values have been between 10% and 50% COHb, there is indication of inhalation of some smoke, which could have contributed to death, or at very least, that the individual was alive when the fire begun. Values below 10% COHb could be suggestive of the fact that the individual was dead prior to fire or died very soon after the fire started (Ellenhorn and Barceloux 1988). However, old people may die at relatively low concentrations, such as 30% COHb. Infants also seem to die at relatively low levels. The variation in fatal concentrations is wide and quite irregular. Any disease process can contribute to death at lower COHb saturation, such as coronary artery disease, respiratory insufficiency, and other debilitating conditions may allow death to occur earlier, when the toxic level is still relatively low. In addition to disease, any separate toxic state is additive (e.g., hypnotic drugs or alcohol) (Saukko and Knight 2004). It has been stated that an accuracy of 10% of the true COHb value is good enough for most cases related to fatalities, as more exact results would not add anything to investigation (Saukko and Knight 2004). However, in order to properly interpret the toxicological findings, it is important for the toxicologist to be aware of the capabilities and limitations of the method that is used to quantitate the COHb saturation levels because they might cause implications in subsequent criminal/or civil litigation.

In the clinical field, on the other hand, diagnosis and treatment of acute CO poisoning still made necessary the existence and usage of rapid and reliable methods for the determination of COHb in human blood. The importance of such methods has been considerably increased by the discovery that COHb fractions of 5% to 10% may aggravate preexisting heart disease (Allred et al. 1991, 1999; Hirchl et al. 2004). In clinical cases, it is important for the applied method to be capable of differentiating between normal (<1.5% for healthy adults and <1% for neonates) and slightly increased concentrations of COHb (e.g., in hemolytic anemia of newborns, or environmental health assessments), in addition to the determination of exogenous CO inhalation which could result in COHb >5% (acute or chronic exposure to CO). In the aforementioned cases (exposure assessment and low level health effects of CO) more valid are methods with high accuracy in the low level range (<5%).

Numerous methods have been developed for the determination of CO in blood, breath, or air. These methods have been ranging from instrumental methods (automated or not) to very simple, semiquantitative or quantitative, manual methods using detector tubes. The developed methods have been based on colorimetry (Maehly 1962; Lambert, Tschorn, and Hamlin 1972; Moffat 1986; Contostavlos and Lichtenwahner 2003), on nondispersive infrared spectroscopy (Morri et al. 1994; Zegdi et al. 2003), on spectrophotometry (Freireich and Landau 1971; Sanderson, Sotheran, and Stattersfield 1978; Brown 1980; Moffat, 1986; Matsuoka 1997; Stonek et al. 2004), and on gas chromatography (Klimisch, Meibner, and Wermicke 1974; Bergman, Coleman, and Erans 1975; Baretta et al. 1978; Griffin 1979; Hobbs, Jachimezyk, and Schloegel 1980; Guillot, Weber, and Savoie 1981; Vreman, Kwong, and Stevenson 1984; Katsumata, Sato, and Yada 1985; Costantino, Park, and Caplan 1986; Goldbaum, Chace, and Lappa 1986; Horvath et al. 1988; Van Dam, Pharm, and Daenes 1994; Kaminski et al. 2003).

We have been running a new Department of Forensic Medicine and Toxicology that would serve to investigate all obvious or uncertain sudden deaths in a restricted area of about 330,000 inhabitants; therefore, it was essential to develop all the methodologies to run the Toxicology Laboratory. We have been wondering which method could be our choice to analyze COHb in blood. The relative literature has been reviewed considering statistical and analytical matters of the most popular and widely used methods for the analysis of COHb in postmortem blood. A critical discussion of these methods has been performed in order to propose the best method for the determination of COHb in postmortem blood that could be utilized in forensic toxicology laboratories where such analyses are limited in number (less than 20 per year). Validation of the discussed methods has been based on previously reported comparative studies (Dijkhuizen et al. 1977; Baretta et al. 1978; Dennis and Valeri 1980; Katsumata et al. 1982; Zwart et al. 1984; Vreman, Kwong, and Stevenson 1984; Goldbaum, Chace, and Lappa 1986; Costantino, Park, and Caplan 1986; Vreman, Stevenson, and Zwart 1987; Fechner and Gee 1989; Mahoney et al. 1993; Van Dam, Pharm, and Daenes 1994; Levine et al. 1996; Canfield et al. 1999; Oritani et al. 2000). Criteria for the evaluation have been considered simplicity, accuracy and precision of results, as well as time and cost of analysis.

METHODS FOR COHb DETERMINATION

The methods firstly developed were colorimetric (Maehly 1962) since the relation of the pink hypostasis with CO poisoning was very early observed. Due to their simplicity and the possibility to export semiquantitative results, these methods (modified or improved) have been continuously used in forensic laboratories in order to export mainly qualitative results (Lambert, Tschorn, and Hamlin 1972; Moffat 1986; Contostavlos and Lichtenwahne, 2003).

However, the most popular and widely used methods have been the spectrophotometric and gas chromatographic ones. The spectrophotometric methods have been performed either by the use of a conventional spectrophotometer with a simple two-wavelength procedure (Moffat 1986; Dijkhuizen et al. 1977; Katsumata et al. 1982; Fechrer and Gee 1989; Canfield et al. 1999), or by dedicated spectrophotometers, which assure, together with the ease and speed of operation, the possibility of measuring other parameters simultaneously (hemoglobin and oxyhemoglobin) (Brown 1980; Dennis and Valeri 1980; Mahoney et al. 1993; Brunelle et al. 1996). Gas chromatography, on the other hand, has been considered an elegant precise procedure, not affected by the quality of the blood tested, and has been considered as the reference method for every CO determination (Costantino, Park, and Caplan 1986; Mahoney et al. 1993; Levine et al. 1996; Oritani et al. 2000).

SPECTROPHOTOMETRIC DETERMINATION OF COHb

Methods for quantifying individual hemoglobin derivatives, as well as total hemoglobin, by the application of the principles of the Lambert-Beer law, have been used both on untreated whole blood and on blood mixed with chemicals to form a stable chromophore since the earliest days of laboratory medicine. Rapid, direct, photometric quantitation of the derivatives, necessary in the clinical environment, has relied on the specific light absorption characteristics of each hemoglobin derivative at the selected wavelengths, which in turn has required independent and exact knowledge of the concentrations of each entity in reference materials.

Two dominant spectrophotometric methods have been commonly performed for the determination of COHb: (i) a manual two-wavelength method (Maehly 1962; Moffat 1986; Katsumata et al. 1982; Fechrer and Gee 1989; Canfield et al. 1999) using simple spectrophotometers; and (ii) an automated multi-wavelength method using specialized spectrophotometers that utilize a set of fixed wavelengths (called CO-oximeters) (Brown 1980; Dennis and Valeri 1980; Mahoney et al. 1993; Brunelle et al. 1996; Brehmer and Iten 2003). It is generally considered that the spectrophotometric methods are limited in sensitivity to approximately 1% COHb of the range of true values (WHO 1999).

Manual Two-Wavelength Method

The easiest and simplest procedure for the determination of COHb in whole blood has been the spectrophotometric two-wavelength procedure, which could be carried out with an ordinary spectrophotometer (Moffat 1986). The method has been sustained in many adaptations and modifications; however, its principle remains substantially the same.

The underlying principle has been quite simple: the spectroscopic methods have made use of the fact that different hemoglobin derivatives (e.g., reduced [HHb], oxygenated [O₂Hb], methemoglobin [MetHb], etc.) have different absorption spectra (Moffat 1986; Zwart et al. 1981). The blood sample should be converted into a two-component system (usually COHb/HHb) and analyzed by measuring at the two different wavelengths selected. The most important considerations for choosing the two wavelengths have been the influence of other dyshemoglobins and of plasma constituents on the absorbance spectra. Both oxyhemoglobin and methemoglobin should be quantitatively converted to the reduced form by the addition in the blood of a reducing agent, whereas COHb could not be reduced. Measurements performed before and after the reduction of hemoglobin at selected wavelengths and the use of relatively simple equations could calculate the percentage carbon monoxide saturation of a blood sample. The result has been expressed as the percentage saturation, meaning the ratio of COHb to total hemoglobin ×100 (% COHb). Methodology has been adapted so as blood samples as small as 10 μl, which could be obtained from a finger prick sample, could be analyzed simultaneously (Katsumata et al. 1982).

This method has been resting on two assumptions: (a) that the samples used for calibration have contained pure COHb and pure HHb, and (b) that the equation(s) used for calculating CO saturation have been valid over the entire range of 0% to 100%. The first assumption has not always been true and could be problematic when attempting to interpret results (Blackmore 1970). Stable and reliable standards should be developed for COHb analysis in postmortem whole blood by procedures difficult to use while the assumed concentrations could not easily be confirmed (Ocak, Valentour, and Blanke 1985; Canfield et al. 1999). The second assumption has been also meeting limitations because it required the exact knowledge of the content of the different forms of hemoglobin present in the sample (Mahoney et al. 1993; Brunelle et al. 1996; Levine et al. 1996). For these reasons, the techniques using two-wavelength measurements would not be expected to be as precise, accurate, or specific as the reference methods, although relative studies support their validity (Table 1).

Automated CO-Oximeters

However, there have been specialized, automated, commercial spectrophotometers, called CO-oximeters, which could provide differential spectrophotometric measurements on hemolyzed blood samples (usually 0.1 ml). These CO-oximeters require little or no sample preparation, provide speed and ease of operation, and can measure total Hb (tHb), COHb, and other forms of hemoglobin simultaneously, in less than a minute (Brown 1980; Brunelle et al. 1996; Zwart et al. 1981; Brehmer and Iten 2003; Lee et al. 2003). The signal is saved, processed, and displayed in a digital form as hemoglobin (g/100 ml) and percentage of oxyhemoglobin and COHb.

Numerous substances and conditions have been found to adversely affect COHb measurements by CO-oximeters. Fetal hemoglobin, temperature, and bilirubin could disturb spectrophotometric COHb determination (Zwart et al. 1981; Steinke and Shepherd 1992; Lambert and Brandt 1993). Methemoglobin, sulfhemoglobin, and oxyhemoglobin have the potential of adversely affecting all of the displayed results of a multicomponent CO-oximeter analysis (Mahoney et al. 1993). Metallic ions and organic compounds, which could adversely affect infrared spectrophotometric measurements of COHb, might also play a role in visible light spectrophotometry. Moreover, errors might arise due to turbidity of the samples containing substantial amounts of lipids (hyperlipidemia) in clinical cases (Hodgkin and Chan 1975; Mahoney et al. 1993). Because CO-oximeters are highly automated instruments, they do not offer the ability of adaptation of wavelengths to particular values suitable for each sample, ability that the manual spectrophotometric method (by the use of a conventional spectrophotometer) could offer (Zwart et al. 1984).

Regarding forensic postmortem materials, the presence of microcoagulates, putrefaction, and contamination could be a significant source of error (Yukawa et al. 1998). It has been stated by many authors that postmortem blood samples had to be pretreated before the COHb determination with a CO-oximeter (Siggaard-Andersen, Norgaard-Pedersen, and Rem 1972; Pannell, Thomson, and Wilkinson 1981; Oritani et al. 1996). The pretreatment has been consisted of reconstitution of blood using detergents and reducing agents, or simply by filtering blood samples to remove any debris and trace clots. However, in such cases doubts about the result should not be excluded.

The listed limit of accuracy by the manufacturers for all these instruments is 1% COHb. Moreover, it has been suggested that no manufacturer could guarantee their accuracy for the analysis of COHb in forensic samples (Yukawa et al. 1998), although different opinions have been also expressed (Levine et al. 1996; Brehmer and Iten 2003; Lee et al. 2003). The precision of measurement with CO-oximeters is excellent, which has misled users regarding their accuracy (Table 1). It looks very reasonable; after all, the primary use of CO-oximeters should be the accurate determination of hemoglobin derivatives in living subjects (patients) acutely poisoned with CO, whereas at low levels of COHb the accuracy is limited.

GAS CHROMATOGRAPHIC COHb DETERMINATION

The gas chromatographic (GC) methods for measuring COHb have been persisted of mixing blood sample with a solution to lyse erythrocytes and then of liberation of hemoglobin-bound CO from blood (Vreman, Kwong, and Stevenson 1984). CO liberation could be achieved through acidification. The released CO has been directed to the reactor headspace, where the headspace gas has been analyzed by GC, as a means of separating carbon monoxide (CO) from other gases. Several different types of detection systems have been applied. Detection has been performed by the use of either flame ionization detector (FID) after catalytic reduction of CO to CH₄ (Griffin 1979; Guillot, Weber, and Savoie 1981; Katsumata, Sato, and Yada 1985; Costantino, Park, and Caplan 1986), or thermal conductivity detector (TCD) (Van Dam, Pharm, and Daenes 1994; Horvath et al. 1988; Goldbaum, Chace, and Lappa 1986), by the use of infrared detection (Maas, Hamerlink, and de Leeuw 1970), by the release of mercury vapor resulting from the combination of CO with mercuric oxide (Vreman, Kwong, and Stevenson 1984), or even by mass spectrometry (MS) (Oritani et al. 2000).

In order to calculate COHb saturation (% COHb), the concentration of total blood hemoglobin (g/100 ml) should be determined with another method. Calibration standards should be prepared by using fresh human blood initially containing <1% COHb (Guilot, Weber, and Savoie 1981). The results have been reproducible and the calibration curve has been linear over the range 0% to 100% COHb. The minimum amount of detectable COHb could be determined by the GC method (Table 1). The reported detection limit for most GC methods has been <0.1% COHb, which is much lower than the normal levels detected for nonsmokers (these were between 0.5 and 1% COHb) (Guillot, Weber, and Savoie 1981). It has been detected even an amount of 0.1 nl of CO in 2 μl of blood with a conventional GC system (FID), which corresponds to a blood saturation of about 0.005% COHb (Vreman, Kwong, and Stevenson 1984).

The GC method has been free from both spectral and chemical interference, which could be caused by contaminants, putrefaction, and the various forms of hemoglobin (Levine et al. 1996). It would yield accurate results on samples exhibiting a high degree of decomposition, provided that a normal hemoglobin level has been assumed. It could be also applied to epidemiological studies of population exposure to low levels of carbon monoxide. The required sample volume could be very small (even 2 μl of blood), thus GC has been particularly suitable for use with blood samples from infants (Vreman, Kwong, and Stevenson 1984). Other studies also have indicated that samples intended for COHb analysis by GC could be stable under most common conditions of transport and storage (Vreman, Stevenson, and Zwart 1987). Regarding forensic materials (blood, thermocoagulated and putrefied blood, or blood containing materials, e.g., bone marrow), GC on molecular sieves with a suitable detection system has probably been the most satisfactory procedure for the measurement of CO liberated from COHb. As it is shown in Table 1, COHb measurements with certain versions of GC have had excellent sensitivity COHb (Vreman, Kwong, and Stevenson 1984; Mahoney et al. 1993), precision (Guillot, Weber, and Savoie 1981; Mahoney et al. 1993), and linearity over a wide range (0% to 100% COHb) of concentration (Goldbaum, Chace, and Lappa 1986; Costantino, Park, and Caplan 1986; Vreman, Stevenson, and Zwart 1987). Moreover, a gas chromatograph with either detector or column could have a broader application than measurement of COHb content only (e.g., detection of other gases, as well as, CO in air). However, for routine or emergency analyses in the clinical field, GC should be considered too demanding and complicated (compared to CO-oximeters) because it requires special trained personnel for proper operation. Moreover, in the forensic field, if large number of samples should be analyzed, the method is considered time-consuming compared to spectrophotometry, while offering a sensitivity that is not usually required (see discussion earlier).

OVERALL ASSESSMENT AND CONCLUSIONS

In conclusion, the selection of the COHb determination method should be made on the basis of its intended use (e.g., quality of sample, number of samples) and its compatibility with other laboratory instrumentation. A forensic toxicology laboratory optimally should have had all the necessary instrumentation in order to detect all the different possible poisons, or at very least it would be acceptable to have the ability to detect the most common poisons (CO is one of them). Spectrophotometers have been considered very basic instruments utilized in every toxicology laboratory and used for many different analyses in the forensic toxicology field.

Listed in Table 2 are the main advantages and disadvantages of the previously discussed methods in respect to analysis of COHb in postmortem blood: the spectrophotometric and the gas chromatographic method.

GC has been more accurate and has been considered the reference method for COHb analysis by many scientists (Costantino, Park, and Caplan 1986; Mahoney et al. 1993; Levine et al. 1996). This is because, unlike spectroscopy, it has been highly specific for CO (Vreman, Kwong, and Stevenson 1984) and could not be affected by substances known to cause spectral interference (Guillot, Weber, and Savoie 1981). It is the most sensitive and precise of the methods discussed. However, GC has been generally limited to research laboratories, because it has been considered relatively complex and time-consuming and required specialized skills for proper operation. In addition, a separate measurement has been required in order to express COHb as a percentage of the total hemoglobin (tHb) in the sample (Guillot, Weber, and Savoie 1981; Vreman, Kwong, and Stevenson 1984; Costantino, Park, and Caplan 1986; Johansson and Wollmer, 1989). It should be mentioned, however, that a gas chromatograph with either detector could be used for much more analyses than only COHb in a laboratory equipped with such an instrument. If variable exposures to CO should be measured (especially low ones) for research purposes, in addition with the determinations routinely performed, then the performance of a GC method on a chromatograph that would also have other applications should be preferred.

The use of the fully automated systems, the CO-oximeters, might offer the most preferable conditions in respect to simplicity. The spectrophotometric COHb inaccuracy claimed by the CO-oximeter manufacturers (in general 1.0%) most likely meets the needs of most clinicians for diagnostic determinations of COHb >5% (e.g., exogenous CO exposure), as well as most forensic samples (Widdop 2002; Saukko and Knight 2004). They have been in use in almost all clinical laboratories where determination of COHb is being performed in patients who have been suspected to be acutely or chronically exposed to CO. CO-oximetry has been the most widely used methodology in forensic toxicology laboratories due mainly to its ease of operation (Brown 1980; Lee et al. 2003) and secondly to the fact that it has been shown to have practical use for analysis of cadaveric blood samples (Lee et al. 2003). However, we are of opinion that the exclusive use of an instrument (although its remarkable simplicity) for the determination of only one substance could be luxurious for a small, new lab if the number of samples to be analyzed per year is not large enough to demand automation. In such a case, the supply of a CO-oximeter could not be considered necessary, because a conventional spectrophotometer with a manual, selected multiwavelength measurement of the critical hemoglobins concentrations would supply the desirable answer.

The manual spectrophotometric method, on the other hand, has been simple to perform and fulfills the accuracy and precision criteria for most analysis of COHb in postmortem blood in assistance with the autopsy findings (acceptable uncertainty 10%) (Saukko and Knight 2004). The use of a conventional spectrophotometer could be considered one of the main advantages of the method, because this has been a basic instrument for any analytical laboratory, and a large number of other analytes could be detected and measured by it as well. Moreover, adaptations in selected wavelengths could be made in order to measure other forms of hemoglobin simultaneously (Zwart et al. 1984).