Comment on ‘Comprehensive toxicological screening of common drugs of abuse, new psychoactive substances, and cannabinoids in blood using supported liquid extraction and liquid chromatography–quadrupole time-of-flight mass spectrometry’

To the Editor, The authors would like to address and clarify information presented in our previous publication [1], as it relates to draft and published standards that were in effect at the time the method was developed. The publication described a supported liquid extraction (SLE) and high-resolution mass spectrometry (HRMS) screening assay for 209 drugs in blood that was intended for drug-impaired driving investigations. The authors use this as an opportunity to (i) clarify differences between recommendations and draft standards that were in place at the time the method was developed and those that currently exist, and (ii) highlight opportunities for additional guidance as standards naturally evolve. The development of standards is a dynamic process. These documents are intended to incrementally improve and inform the forensic science community on best practices and minimum requirements, while being responsive to technological advances and innovation. Standards are made publicly available in draft form during the public commenting period (before final approval) and are subject to regular updates on a cycle determined by the standards developing organization (SDO). Likewise, recommendations for the toxicological investigation of drug-impaired driving have evolved considerably since they were first published in 2007 [2]. At the time that our HRMS screening method was developed, the 2017 recommendations were the most current [3] and ANSI/ASB 120 [4] had not yet been published. However, a draft first edition was available. Therefore, the cutoff concentrations that were selected for the method were based upon the 2017 published recommendations [3] and the 2019 draft version of ANSI/ASB 120. As a result, changes to the draft standard that took place after our method development are not reflected in the publication [1]. The authors regret not making it clear in the article that cutoff concentrations adhered to the draft standard (available at the time), not the final published version (after validation). Therefore, although the publication described a method that certainly exceeded the minimum requirements in terms of analytical scope, it did not meet analytical sensitivity requirements for certain analytes according to the final published standard. Regrettably, the authors were not aware of the impact of the changes. The authors take this opportunity to highlight those differences and discuss challenges moving forward. When the screening cutoffs for immunoassay (IA) and non-IA based testing were removed from the 2019 draft of ANSI/ASB 120, a footnote was added to the final approved standard that directs laboratories to apply blood confirmation cutoffs in lieu of screening cutoffs (if immunoassay is not used). As a result, the cutoffs used during our method development and validation are different for nine compounds in ANSI/ASB 120, notably buprenorphine, 11-nor-carboxy-Δ9-tetrahydrocannabinol (carboxy-THC), diazepam, fentanyl, methadone, nordiazepam, oxazepam, temazepam, and tramadol. Table 1 highlights these differences and compares the cutoff concentrations utilized in our study [1] with the recommendations in place at the time of development [3] and at present [5], with those in the draft (at time of development) and final version of the standard [4]. The application of blood confirmation cutoffs for non-immunoassay screening in ANSI/ASB 120 effectively lowers the cutoff concentration for these drugs (Table 1). Differences in cutoff concentrations between the published article, recommendations, and standards High-dose benzodiazepines. Differences in cutoff concentrations between the published article, recommendations, and standards High-dose benzodiazepines. Although these were not the cutoff concentrations used in our study, the limits of detection (LOD) for the high dose benzodiazepines, methadone, and tramadol are sufficiently low for this not to create a performance issue (Table 1). However, the changes in cutoff concentration for fentanyl, buprenorphine, and carboxy-THC are likely to present a challenge. Although the LOD for buprenorphine (0.25 ng/mL) was slightly lower than the confirmation cutoff, it was equivalent for carboxy-THC (5 ng/mL) and higher for fentanyl (1 ng/mL). Therefore, the authors wish to make it clear that while the published method [1] met sensitivity requirements for immunoassay-based screening, the increased sensitivity required to achieve confirmatory cutoffs might not be possible for three of the compounds in Table 1. The published method [1] was validated with guidance from ANSI/ASB 036 [6], which is currently undergoing revision. This standard and the 2013 guideline that preceded it [7] provided the foundation that has guided method validation in forensic toxicology for over a decade. The 2019 version of ANSI/ASB 036 describes minimum validation parameters for IA-based and alternative targeted screening methods. However, at the time these validation documents were conceptualized, mass spectrometry (MS) based screening was significantly less widespread. Moreover, the array of modalities which are now available, including the ability to perform both targeted and non-targeted MS-based screening, have created somewhat of a paradigm shift [8, 9]. Future versions of standards that address validation, identification, and mass spectral acceptance will undoubtedly consider these technological advances, reinforcing the very nature of the standard environment, which is intended to constantly evolve. While operational forensic laboratories use screening techniques to rule out substances or indicate further testing, the increased technological capability of HRMS for toxicological screening offers significant opportunities and safeguards to improve confidence in a presumptive positive result. However, the myriad of acquisition modes for targeted and non-targeted screening can also present challenges in terms of workflow, analytics, and efficiency. Whole blood (bovine) containing 1% sodium fluoride and 0.2% potassium oxalate was used for all method development and validation experiments described in our article [1]. The practice of using non-human blood for method development, validation, and quality controls is routine in operational forensic laboratories. If validation of a method requires matrix (i.e. blood, urine, vitreous, tissue) from a particular genus or species, this would be specified in the standard. Current standards for method development and validation in forensic toxicology [6] do not preclude the use of non-human blood for the development and validation of methods intended for whole blood in humans. Although described elsewhere, we also analyzed >900 previously adjudicated human blood specimens to compare presumptive positive IA and MS-based screening results [10]. In its current form, ANSI/ASB 036 requires non-IA screening methods to include interferences studies, LOD, matrix effects (ionization suppression/enhancement), and processed sample stability (if applicable). Future versions of the standard might include additional parameters (e.g. performance or precision at the decision point), or guidance on how to approach the evaluation of these parameters. For example, how should operational laboratories approach the evaluation of ionization suppression or enhancement for screening methods involving hundreds of analytes? The current standard [6] allows significant flexibility, allowing for both dynamic (post-column infusion) and static (post-extraction addition) methods. While the authors of the present study utilized both post-column infusion for all analytes and internal standards, and quantitatively assessed level I analytes using post-extraction addition, both are currently valid approaches under ANSI/ASB 036 [6]. Future recommendations or standards may address the appropriateness of these or consider alternative ways to approach the evaluation of ionization or suppression for screening or confirmatory use. Although we evaluated ionization suppression and enhancement for all 209 compounds and 10 internal standards, this was a major undertaking. As a compromise, we elected to perform this only at the most challenging (low) concentration of analyte (i.e. decision point) and not at the high concentration. When ionization or suppression criteria are not met, ANSI/ASB 036 requires that the laboratory demonstrate that there is no impact on other critical validation parameters [6]. While the most critical validation parameter for a qualitative method might be the LOD, for a presumptive drug screen the critical concentration is most likely considered to be the decision point or cutoff. As an example, ANSI/ASB 036 describes tripling the number of blank matrices evaluated for LOD to nine when ionization/suppression criteria are not met. In our study, ten independently sourced blood specimens were evaluated at the decision point. These were described in the article, and performance at the decision point for all 209 drugs was assessed, including reproducibility between matrices. This included aged and degraded blood that had been stored for ≥5 years. As ANSI/ASB 036 undergoes revision, future editions might provide additional guidance on reasonable approaches to assess “critical validation parameters” for various types of testing when ionization/suppression criteria are not met. The informative annexes that are used in the current standard [6] include examples that are extremely helpful and can aid in the interpretation of the standard itself. While all standards are intended to be unambiguous, it is inevitable that some provisions will be interpreted differently. This again speaks to the iterative nature of standards, and the opportunity to improve and clarify requirements with each edition. Although our SLE development and screening method included substances outside of impaired driving [i.e. non-steroidal anti-inflammatory agents (NSAIDS), salicylates, and non-narcotic analgesics (e.g. acetaminophen)], the compounds were included to increase and challenge the potential scope of SLE, since many laboratories perform testing for medicolegal and impaired driving purposes. As indicated in the article, the method was validated using blood only. Although analytes of interest that might be encountered in urine were included for some drug categories, no evaluations were performed on matrices other than whole blood. The current version of ANSI/ASB 036 provides multiple approaches to LOD for immunoassay screening, non-instrumental, and qualitative methods. While LOD determinations for qualitative or quantitative instrumental methods are minimally determined in three sources of matrix (in duplicate) over three runs, immunoassay screening allows for a single source of fortified matrix. In practice, many laboratories will elect to assign the LOD administratively as the decision point, and this is permissible for immunoassay screening [6]. As stated in our study, an empirical approach to LOD performance was undertaken. Blood specimens for screening were pooled, rather than assessed separately (n = 3). As a result, the reproducibility at the critical decision point was further evaluated as described above using ten independent sources of blood. In our study, LODs were not established unless all applicable criteria were met, including signal to noise ratio (SNR), retention time (RT), mass accuracy, ion ratio, and algorithmic (coelution) score. These criteria highlight that targeted and non-targeted MS screening provide criteria for presumptive identification that far exceed the capabilities of immunoassay. While this increases confidence in the results, the added burden of identification criteria can impact workflow. While MS-based technologies advance screening capabilities and confidence, robust studies to address analytical benefits and overall workflow efficiency are greatly needed. Although ANSI/ASB Standard 098 [11] was not published at the time of method development, drafts of this standard only required a minimum of a single diagnostic ion for multi-stage HRMS screening. The final published standards related to mass spectral analysis and identification criteria now take a completely different approach: ANSI/ASB 098 [11] no longer recommends minimum requirements for screening and confirmation. Instead, it is designed to be used in conjunction with ANSI/ASB 113 [12] to identify an analyte based upon a point system. Multiple stringent criteria were utilized in our screening assay, including retention time, signal to noise ratio for both precursor and fragment ions, mass accuracy for both precursor and fragment ions, ion ratios, and coelution scores. Fragmentation was analyte dependent and 22 of the 209 drugs did not produce quality fragments. For those analytes, RT, SNR and mass accuracy were evaluated. All Ions data acquisition and MRM data analysis were performed using Agilent MassHunter Quantitative software as previously indicated [1]. Isotopic distributions were not selected using the data analysis workflow, but these could have been included. As discussed, the criteria far exceed those typically required for toxicological screening, and laboratories should select the critical parameters that are fit for purpose, align with their overall analytical scheme, and maximize efficiency. While HRMS screening offers a vast array of possibilities in terms of the data available for critical evaluation, it can also depend on the instrument (hardware), software, and the acquisition mode that is selected. Future editions of standards related to mass spectral analysis will continue to address these issues. Due to the rapidly changing pace and acceleration of technologies, these particular standards may require more frequent updates. In discussing our approach, we have tried to highlight additional limitations, practical considerations, and opportunities for future standards to provide guidance as the array of targeted and non-targeted MS-based screening technologies evolve. Just as our method development, optimization and validation of the screening assay was undertaken over several years, consensus-based standards also take years to develop. As a result, the criteria selected for method performance may change. Here we clarify that adherence to standards in our published method [1] was based on documents that were available at the time of development. The authors regret not placing greater emphasis on the versioning of standards in the original article, but have used this as an opportunity to highlight some of the differences and increase overall awareness of the dynamic nature of the standards environment. None declared.