An Indisputable Case of Matrix Effects in Blood Alcohol Determinations

Sep 07, 2016

In a recent review of blood alcohol casework performed by a forensics laboratory associated with a major metropolitan police force, I was again (1) disheartened to find major deficiencies in method validation protocols. In this case, the analysts failed to check whether aqueous solutions for calibration and quality control were reliable surrogates for real blood samples. What I describe here is the definite possibility that matrix effects have caused this laboratory to overreport blood alcohol concentrations (BACs) determined on one of their headspace gas chromatography (HS-GC) systems since 2011. The fact of the matter is that this forensics laboratory would never be able to dispute this claim, because they lack the protocols and data that would be necessary to check for such an effect. Importantly, best practices and trusted guidelines say they should have checked for this effect as part of a comprehensive ongoing method validation and revalidation plan, but they have failed to do so.

An internal standard HS-GC method was used to measure BAC. As is generally the case, n-propanol was used for the internal standard, since it is unlikely to be found naturally in human blood samples, it has similar properties to the target analyte, ethanol, and it produces a distinct signal for measurement (that is, a separate peak). n-Propanol is added in a consistent quantity to every calibration standard, quality control, and real case sample. It is then measured along with the ethanol during each analysis. This measurement is done to normalize ethanol responses and reduce systematic and random errors associated with sample preparation, injection, chromatographic separation, and detection. A key point is that the internal standard should behave essentially identically to the analyte throughout these steps, so that any losses or gains experienced by the analyte would be also experienced by the internal standard, and therefore corrected in the final unknown determination.

In the case (or cases) in question, data indicated it is plausible that a matrix effect altered the response of the internal standard. Matrix effects are known to systematically alter reported results, if they are not accounted for and corrected. They can be sample dependent, analyte dependent, and concentration dependent. A good analytical scientist will always seek to either use a matrix for calibration that is essentially equivalent to the samples tested (that is, prepare spiked standards into a blood matrix), or conduct validation measurements to check whether calibration in a surrogate matrix (that is, an aqueous solution) is valid for determinations in samples of a different matrix. Neither of these steps were performed by the forensics lab.

The laboratory’s analysis was performed using a graphical internal standard calibration. A series of calibration solutions was prepared in aqueous solution (a sodium chloride–fortified water solution) in which the solutions contained varying amounts of ethanol and a consistent amount of n-propanol. Each of the standards were analyzed by HS-GC, and the relative responses (response of ethanol/response of n-propanol) of the standard solutions were plotted against the relative amounts (amount of ethanol/amount of n-propanol) to create a calibration curve. Over a relevant range of ethanol levels, a linear equation can be generated for use in determination of unknowns. The unknown solution (blood matrix) was prepared to contain an equivalent amount of internal standard as in the calibration standards. When the sample was analyzed, the relative responses of ethanol and n-propanol, along with the known amount of n-propanol added, allowed the analyst to determine the unknown ethanol concentration using the previously established linear equation. A problem could arise when the components of interest (the ethanol and n-propanol from the aqueous matrix and the ethanol and n-propanol from the blood matrix) are not equivalently transferred from the sample into the chromatographic system.

In this case, it was evident that the transfer efficiency of the n-propanol internal standard into the headspace was disproportionately lowered during the analysis of the real samples, compared to the analysis of standard aqueous samples. Overall, peak areas for the internal standard response were consistently in excess of 20% lower when analyzed from blood samples, compared to when they were measured from standard aqueous samples. From the data I evaluated, this is a trend that appears to be consistent on this particular instrument since 2011. From the data alone, it is hard to surmise the exact mechanism of this matrix effect—whether it is chemical or instrumental in nature.

What is an absolute travesty is that no measurements were ever performed to check whether ethanol responses were similarly or differently affected by the matrix. Well, I suppose the analysts never recognized the systematically low response of the n-propanol, but the main point is that such checks should be built into a reliable method validation plan.

So, if we assume that the matrix only had a response lowering effect on the internal standard and not on the analyte—which is an absolutely possible situation and is indisputable based on the lack of data to show otherwise—then all of the reported ethanol values from blood samples would be high, perhaps more than 20% high. In other words, a lower internal standard response caused by matrix effect, with a consistent ethanol analyte response, would increase the calculated relative response used to determine the unknown concentration. The result would be an artificially high value reported for the unknown analyte based on calibrations performed in an aqueous matrix.

Based on a 20% bias, someone with an actual 0.07-g/dL BAC (below the legal limit) could register a 0.084-g/dL BAC (above the legal limit) according to the assay. And that is just a conservative average. One could imagine, on a per sample basis, much more extreme biases occurring. Again, the point is that the apparent bias was never noticed and evaluated.

There are more nuances to this case, including the use of only a four-point calibration, failure to comprehensively check bias, precision, and carry-over, and lack of revalidation following instrument maintenance. I have been employed as a consultant to review this and other cases. In some cases (based on data from other forensics labs), I find nothing much of note to report. However, this is a case where well accepted and published best practices for method validation have not been followed. As a result, there could be some significant injustice. Rather than addressing these on a case-by-case basis for various defendants, it would seem more efficient to work together with forensics laboratories to ensure their method validation is up to par. However, anecdotally, I am told that the likelihood of such a thing happening is very small, because they are too set in their ways. That is a shame. Rulings of guilt associated with blood alcohol cases and driving under the influence are truly life crippling. Where justified, these decisions should be rendered, but this seems a setting where the laboratory really has to have the analytical science correct. In the case I described above, I believe they do not.

 

Reference

1. K.A. Schug, The LCGC Blog, June 8, 2015. http://www.chromatographyonline.com/lcgc-blog-forensics-lawyers-and-method-validation-surprising-knowledge-gaps

Kevin A. Schug is a Full Professor and Shimadzu Distinguished Professor of Analytical Chemistry in the Department of Chemistry & Biochemistry at The University of Texas (UT) at Arlington. He joined the faculty at UT Arlington in 2005 after completing a Ph.D. in Chemistry at Virginia Tech under the direction of Prof. Harold M. McNair and a post-doctoral fellowship at the University of Vienna under Prof. Wolfgang Lindner. Research in the Schug group spans fundamental and applied areas of separation science and mass spectrometry. Schug was named the LCGCEmerging Leader in Chromatography in 2009 and the 2012 American Chemical Society Division of Analytical Chemistry Young Investigator in Separation Science. He is a fellow of both the U.T. Arlington and U.T. System-Wide Academies of Distinguished Teachers.

lorem ipsum