Organophosphate nerve agents (OPNA) remain a public health threat due to their continued use despite international prohibitions [1, 2]. Dissemination of these compounds through warfare and terrorist activities has resulted in human exposures causing injury or death [3]. These exposures demonstrate the need to identify the exposure agent as well as differentiated from the worried well from persons with low-dose exposures to monitor for delayed health effects. Exposure to OPNAs can be confirmed by the detection of specific biomarkers, including protein adducts and metabolites [4–8]. The alkyl phosphonic acid metabolites, which readily form by hydrolysis in the body, can be used to diagnose OPNA poisonings; these metabolites have been measured in blood products following human exposures to VX and sarin (GB) as well as in animal studies with GB and cyclosarin (GF) [9–12]. Concentrations of the OPNA metabolites are rarely reported; however serum levels of isopropyl methylphosphonic acid (IMPA, metabolite of GB) in victims following chemical weapons attacks in Tokyo and Matsumoto ranged from 2 to 136 ng/mL [10].

Methods to detect and quantify these metabolites in clinical matrices have been established using LC-MS/MS [6, 10, 12–16] and GC-MS [17–19]. Sample preparation continues to be critical for analysis to reduce potential interferences and ionization effects and increase sensitivity for MS detection. Previous studies have focused on isolation of OPNA metabolites from blood and urine matrixes using ion exchange, reverse phase, and normal phase SPE [14, 20, 21]. While these protocols have proven effective, they can be time consuming, costly, and may require a larger sample volume (50–1000 μL) [14, 20, 21].

Rapid sample collection is crucial to identify affected persons, due to decreasing concentrations of metabolites. Additionally, blood concentrations of OPNA metabolites may vary greatly depending on the length of time from exposure to sample collection, the level of exposure to the agents, and individual biological factors [9, 12]. Alternatives to invasive, labor-intensive blood draws are needed to accelerate and simplify the sample collection process. The use of blood spotting cards [22, 23] and microsamplers [24, 25] may be implemented outside of the clinical setting, thus providing suitable alternatives to traditional blood tubes.

Dried blood samples have been used for epidemiological disease surveillance to rapidly collect samples from large numbers of people [26, 27]. There are multiple advantages to using dried blood spots (DBS) over traditional blood collection techniques. Dried blood samples are collected by spotting or sorbing venous blood or blood from a finger prick onto filter cards or microsamplers. Collecting dried blood samples is generally faster and less invasive than a blood draw, uses a smaller volume of blood, and requires minimal training, meaning that self-sampling may be possible. Furthermore, blood spot cards can be maintained at ambient temperature with desiccant and do not require cold storage. Finally, because blood is dried on the collection cards, samples are considered stable and non-hazardous during storage and transport [28, 29].

Impacts resulting from variations in the blood must be considered when performing analytical quantitation from dried blood samples. The most critical factor to address is hematocrit (HCT), a measure of red blood cell (RBC) volume, which can cause spot size variability and heterogeneous distribution of blood components [30]. These two issues may be mitigated by analysis of the entire spot [31]. However, extraction differences due to chemical binding to blood have also been identified in samples with different HCT and must be investigated [28, 30].

For accurate quantitation of OPNA metabolites in a suspected exposure sample, the volume of blood sample must be known [30, 31]. Pipetting can be used to control the spot volume; however, in an emergency, more user-friendly devices such as microsamplers and controlled-volume capillaries may be beneficial. Microsamplers are devices used to obtain dried blood samples for quantitative bioanalysis and have been documented to collect the same volume of blood across HCT variations without the need for additional volume measurement [24]. Because microsamplers are polymeric instead of cellulose, differences in extraction from each sorbent could be possible. For these reason, microsamplers were evaluated along with traditional DBS in this study.

In some cases, microsamplers may not be available and pre-measurement of the blood volume prior to spotting may not be feasible. These samples can still be used to identify exposure to OPNAs; however, accurate quantitation may be challenging. Comparison of metabolite concentrations to previous exposures is desired to better interpret the exposure event. A previous study normalized OPNA protein adduct concentration with total protein content in DBS to achieve this goal. Through this correction, a more comparative value for diagnostic interpretation was achieved [5].

For the first time a novel method has been developed to quantitate metabolites of VX, GB, soman (GD), GF, and VR in dried blood using analysis by hydrophilic interaction liquid chromatography (HILIC) coupled with tandem mass spectrometry (MS/MS). The method was evaluated for precision, accuracy, sensitivity, matrix effects, and extraction recovery. Effects of hematocrit on quantitation were investigated for DBS and microsampling devices. Additionally, the normalization of metabolite to protein content was studied to support diagnostic interpretation of results from DBS without controlled volume collection. This application of DBS can support the rapid collection of exposure samples following an emergency event.

Various dried blood samples were evaluated for the determination of OPNA metabolites in whole blood. An extraction of 5 L DBS and 10 μL dried blood microsamplers was developed and validated for the quantitation of these five metabolites using LC-MS/MS. Since samples received following an emergency or large-scale exposure event may vary in spot volume, the normalization of OPNA metabolite concentration to total protein concentration was also investigated to correlate results from DBS with uncontrolled volumes for diagnostic interpretation.

In initial experiments, 5 μL DBS were extracted with 5% DI water in acetonitrile to align the extraction solvent with the established HILIC chromatographic separation and avoid the need for additional sample preparation steps [14, 33]. Using this extraction, OPNA metabolites could not be detected at 50 ng mL⁻¹; therefore, a more polar solvent and an additional concentration step were investigated to increase extraction efficiency and improve sensitivity. DI water and methanol were selected for extraction evaluation. The extraction with DI water successfully removed the OPNA metabolites from the blood card, but additional time was required to concentrate the extract. The use of methanol was ultimately selected as it successfully extracted the metabolites with the least concentration time needed. The volume of solvent used for extraction was also investigated. Minimal differences in response and reproducibility were noted between 1000 μL, 500 μL, and 250 μL of extraction solvent; therefore, a volume of 250 μL of methanol was selected. These extraction parameters were applied to a 30 ng mL⁻¹ DBS sample and analyzed by HILIC LC-MS/MS (Figure 1). Contamination of the single-hole punch was also investigated. For the blank paper samples and blank DBS samples punched immediately following the highest calibrator, no responses were detected.

DBS were initially extracted in glass vials using sonication; however, this process was labor and material intensive. Incorporation of 96-well plates permitted the use of multichannel pipettes for solvent addition and transfer steps, reducing sample preparation time. Use of this container was anticipated to impact DBS extraction; therefore, mixing time was optimized for the 96-well plate format to ensure the optimum recovery of the OPNA metabolites. Mixing times of 5, 10, and 15 min were evaluated, with a 10 min mixing selected since the response and reproducibility increased from 5 to 10 min, but no further increases were observed at 15 min. The final extraction recoveries from DBS in 96-well plates were comparable to those obtained in individual glass vials. Following the success of this process for DBS, microsamplers were prepared in the same manner.

With this protocol, extraction recoveries were determined at 30 and 250 ng mL⁻¹ for both DBS and microsamplers using pooled whole blood (Table 1). For microsamplers, the extraction recoveries were slightly lower for the more polar metabolites, IMPA (87.9%) and EMPA (85.0%); this difference was not observed for DBS. The remaining metabolites, PMPA, CMPA, and MMPA, were extracted from the spots and microsamplers with recoveries all above 95.2%, indicating a near complete recovery of analyte.

Matrix effects, evaluated at 30 and 250 ng mL⁻¹, ranged from 26.1–33.3% for DBS and from 60.3–83.3% for microsamplers for all five analytes. The absence of matrix effects was defined as zero; therefore, the microsamplers had almost double the matrix effects as compared to the DBS. An assessment of matrix effects for comparable volumes, using two DBS spots equaling 10 μL of sample, did not indicate that volume was a significant contributing factor for matrix effects. Therefore, this variation in matrix effects between the sample collection types was likely a difference in retention of blood components to the blood spot cards (composed of cellulose) as compared to the microsampler (composed of polymeric sorbent) [24].

The estimated LOD values (Table 1) were below the lowest calibrator of 5 ng mL⁻¹ for all analytes. Variability at 5 ng mL⁻¹ was 9.8–14.5% and 11.1–16.3% for DBS and microsamplers respectively; mean accuracies at the same level were within 89.6–106% for all five analytes for both sample collection types. This data in combination with the estimated LOD supports the selection of 5 ng mL⁻¹ as the lowest reportable limit for these samples.

Twenty curves for both DBS and microsamplers were prepared and analyzed to determine method accuracy and precision. Multiple analysts participated in curve preparation and analysis, and no more than two curves were analyzed per day. Accuracy of the QC means for all analytes fell between 93.0 – 108% for microsamplers and 93.7–108% for DBS, which were within FDA recommended guidelines of 85 – 115% (Table 2) [34]. Coefficient of variation for the QCs for both DBS and microsamplers ranged from 4.79–16.0%. QC variability was lower overall for DBS (average 8.15%) as compared to the microsamplers (average 13.5%). Challenges preparing the microsampling devices from pre-collected blood occasionally resulted in an excess of sample on the shoulder of the tip, which may have contributed to this increase in bias [24].

Unfortified individual whole blood samples were evaluated for the presence of potentially interfering components. No peaks were detected at corresponding retention times for any of the five OPNA metabolites in 80 individual unexposed blood samples prepared and extracted from DBS. This observation was confirmed for microsamplers using a subset of these individual unexposed samples. Therefore, the potential for false positives from the matrix was determined to be minimal.

OPNA exposure samples are difficult to obtain; therefore, blank individual blood samples were fortified with OPNA metabolites for method assessment. As described before, hematocrit impacts the extraction of analytes from the matrix; therefore, potential biases from variations in HCT were evaluated for DBS and microsamplers. The typical reported ranges of HCT in healthy adults is from 35–50% [35]; higher or lower values may occur when health of an individual is compromised. To evaluate the normal HCT range, individual blood samples (HCT values 25.8–47.0%) were fortified at 30 and 300 ng mL⁻¹ and evaluated as unknowns for both DBS (n=5) and microsamplers (n=4). The mean accuracy for all five metabolites at 300 ng mL⁻¹ ranged from 91.0–114% with precision not exceeding 12.6% for both sample types. At 30 ng mL⁻¹ the mean accuracy ranged from 76.4–100%, with precision not exceeding 7.19% (with the exception of EMPA extracted from microsamplers which had a variability of 22.6%). Overall, these results confirm the effectiveness of this method to extract OPNA metabolites from DBS and microsamplers from individual blood samples within the typical reported HCT range in healthy adults without significant biases or variance.

To investigate the impact of HCT outside the normal range, pooled blood with adjusted HCT ranging from 15–70% was fortified with OPNA metabolites at 30.0 and 300 ng mL⁻¹ and used to prepare DBS and microsamplers. All hematocrit levels were measured prior to fortification with OPNA metabolites and preparation of DBS and microsamplers. No correlation between HCT and recovery was observed for DBS both spike levels (Figure 2). Microsamplers, however, showed a slight bias at lower HCT for EMPA and were strongly correlated with a low bias at high HCT for all analytes. Thus the use of mictosamplers exceeds the method accuracy expectation at hematocrit levels below 25.8 for EMPA and above 47.0 for all analytes tested [34].

In the event of an emergency, samples may be collected in a variety of ways, including DBS of uncontrolled volume. To obtain quantitative data from an unknown volume of blood, normalization to a consistent component of whole blood is needed. A previous study has referenced total protein for normalization of other OPNA biomarkers from DBS [5]. To evaluate this process for OPNA metabolites, 100 μL DBS were prepared using fortified whole blood containing the five nerve agent metabolites. A single punch constituting approximately 40% of the blood volume was extracted with 100% methanol, using the 5 μL DBS protocol described above. Although the metabolites were successfully extracted, the blood proteins fixed irreversibly on the card, and total protein could not be measured. A solution of 100% DI water successfully extracted both the metabolites and proteins from the DBS. An aliquot of the aqueous extract was then used to measure total protein. The remaining aqueous extract was concentrated by evaporation, reconstituted in 5% DI water in acetonitrile, and analyzed by HILIC LC-MS/MS.

DBS of 5 and 100 μL volumes were prepared from a whole blood sample fortified at 200 ng mL⁻¹, extracted with DI water, and analyzed as above. The OPNA metabolites were quantitated relative to a calibration curve prepared from 5 μL DBS extracted with DI water. As expected, the volume difference resulted in a significant bias between the samples (Figure 3A); thus, the analyte concentrations from both 5 μL and 100 μL sub-punch DBS were normalized to total protein content to better compare the results. A relative ratio of metabolite concentration to protein content significantly improved the comparison between the whole and partial spots (Figure 3B). This correlation provides needed information to improve interpretation of OPNA exposures from DBS with unknown blood volumes.

The methods presented here have the ability to detect human exposure to five nerve agents from dried blood samples. Although other methods have reported lower detection limits [6, 13, 36], these unique methods will identify the presence of OPNA metabolites as low as 5 ng/mL. Furthermore, the analysis of a single dried blood sample can be completed in approximately 30 min, providing a rapid response, similar to previous publications [36]. However, the greatest advantage of this analytical approach is the ability to detect and quantitate neve agent exposure from dried blood samples, which will be benifical in the event of a large scale exposure

Dried blood samples may be an effective tool of collecting human exposure samples from a large chemical release. Five nerve agent metabolites were detected from 5 μL DBS and 10 μL microsamplers from 5–500 ng mL⁻¹ with precision <16% and accuracy between 93–108% for QC samples. Evaluation of individual samples confirmed lack of interferences and minimal biases resulting from matrix effects or HCT variations, with the exception of microsamplers for HCT outside of the normal healthy range. Bias observed when evaluating DBS of uncontrolled volumes was substantially reduced by normalizing measured OPNA metabolite concentrations to total protein. This normalized measurement will improve interpretation of nerve agent exposures when using alternative collection techniques such as DBS when volume of blood is not known. The use of microsamplers could facilitate self-sampling supporting the need for rapid collection of clinical samples; however, controlled volume DBS is preferred as filter paper for preparation of spots is readily available and there are minimal quantitation issues related HCT.