High-Precision Quantitation of Biofluid Samples Using Direct Mass Spectrometry Analysis

Abstract

The transition of Mass spectrometry (MS) for clinical analysis is highly desirable and a major progress has been made with direct sampling ionization for operation simplification. High-precision quantitation, however, remains as a major challenge in this transition. Herein, a novel method was developed for direct quantitation of biofluid samples, using extremely simplified procedure for incorporation of internal standards selected against the traditional rules. Slug flow microextraction was used for the development, with conditions predicted by a theoretical model, viz. using internal standards of partition coefficients very different from the analytes and large sample-to-extraction solvent volume ratios. Direct quantitation of drug compounds in urine and blood samples was demonstrated. This development enabled an extremely simplified protocol that is expected to have a significant impact on on-site or clinical analysis.

Graphical Abstract

Biofluid samples, such as urine, blood, or cerebrospinal fluids, are often analyzed for clinical diagnosis.^1,2 As a general purpose technology for chemical analysis, mass spectrometry (MS) is routinely used in analytical laboratories to provide analysis at high sensitivity and high specificity.^3,4 High precision quantitation can also be achieved at low concentration levels using MS⁵ and internal standard (IS) incorporation as well as sample purification are typically necessary to overcome the matrix effects.⁶ Ambient ionization enabled direct MS analysis of complex samples, where direct sampling and direct ionization were integrated for minimizing the overall sample preparation prior to MS analysis.^7–11 Fast extraction-based methods, such as solid-phase microextraction,¹² and polymer coating transfer enrichment¹³ have also been introduced for direct MS analysis. As a prominent trend, the combination of the direct sampling/ionization with miniature mass spectrometers,^14–16 is expected to lead to a feasible solution for future on-site and clinical MS analysis.

Although the outlook for implementing MS analysis with simple procedure appears promising with the recently developments, the quantitation, which is often mandatory for direct and fast analysis, has not been proven to be highly feasible with simplified procedures. Ambient ionization may have only provided half of the solution in terms of minimizing the sample matrix effects; accurate IS incorporation is necessary for high-precision quantitation while its implementation with simple operation, however, remains challenging. In a traditional lab procedure of quantitating analytes in biofluid samples, volumes of all the samples and solvents involved in the sample treatment and IS incorporation need to be accurately controlled. This is difficult to be transferred for on-site or in-filed testing, where lab techniques typically are not feasible to perform. A variety of methods have been explored previously for accurate sampling, such as using capillary volume to control the sampling amounts.¹⁷

In this study, we intended to demonstrate a new strategy to further simplify the analysis procedure by avoiding volume controls in quantitation using direct analysis by mass spectrometry while still retaining high precisions. This might appear to be a trivial step but has been a major barrier preventing ambient mass spectrometry from being applied for quantitative point-of-care (POC) testing, for which lab techniques such as pipetting is not allowed in the procedures. The demonstration was established for analysis of target analytes in biofluid samples, with the sampling technique used falling into the same general category as slug flow microextraction (SFME).^18,19 SFME uses a thin capillary to hold two immiscible liquid plugs of the extraction solvent and the sample in contact. The movement of the plugs in the capillary induced turbulences inside each liquid phase, which facilitates a fast-chemical exchange at the interface. Combined with electrospray ionization (nanoESI) MS, direct analysis of chemicals in blood and urine samples of small volumes (~5 μL) was performed. Limits of detection (LODs) as low as 0.03 ng/mL was obtained for analysis of illicit drugs and their metabolites in urine samples. On-line chemical derivatization could also be implemented to improve the sensitivity for analysis of steroids, which typically do not ionize efficiently through electrospray methods. Three-phase SFME and reverse-phase SFME modes were further developed for analysis of polar compounds in biofluids and vegetable oils, respectively.²⁰ Good quantitative performance was achieved by these SFME-based MS methods; however, accurate control of the volumes for both the samples and the extraction solvents was necessary

For the simplified quantitation procedure with no volume controls for samples or extraction solvents, the developed method was shown to be applicable to quantify polar compounds in relatively large volumes of biofluid samples. To apply the SFME to extract analytes from a liquid sample of relatively bulky volume, a capillary was loaded with a small volume of ethyl acetate (5–10 μL) and then one end was dipped into the bulk urine sample with a direct contact created between the two immiscible phases (Figure 1a, b and Figure S1). The liquid-liquid extraction was facilitated by the slug flow motion of liquid plugs, which is generated by simply adding a push-pull force to the other end of the capillary using a pipette or bulb.

Figure 1.

a) Slug flow microextraction of analytes from a sample of large volume; b) MS/MS spectra of 50 ng/mL verapamil in urine of 5 μL and 2000 μL using slug flow extraction followed by nanoESI MS analysis.

We explored SFME for high-precision quantitative analysis of biofluid samples of relatively bulky volumes (milliliters), which typically are readily available for samples such as urine. Apparently, a sampling of analytes from a relatively large sample volume into an extraction solvent of small quantity can be expected to improve the sensitivity for the analysis, due to a high-volume ratio preconcentration effect. However, we also strived to achieve high precisions in quantitation without requiring the controls that are typically critical for quantitative analysis, such as the accurate measurements of the volumes of the sample or the extraction solvent, or careful mixing of the IS. This represents a significant advantage for transferring MS analysis for POC applications since the operation procedure would be greatly simplified.

The sample plug in the capillary is always connected to the bulk sample for SFME. While the chemical exchange occurring at the interface of the extraction solvent the sample, fresh sample is constantly filled into the capillary from the bulk volume. While performing the SFME, the dipping end of the sampling capillary was also moved around to gently stir the sample, which helped to keep the homogeneity of the sample. After the extraction, the extract solvent was transferred into a borosilicate glass capillary with a pulled tip; and a high voltage was applied through a metal wire to induce nanoESI for MS analysis. The relative chemical distributions in the bi-phase system after the extraction can be estimated based on their partitioning coefficients (logP) when an equilibrium is achieved.¹⁸ When a large volume of sample is provided for the extraction into a solvent of small volume, “unlimited” amount of analytes are provided for the chemical extraction. Therefore, a highest analyte concentration in the extraction solvent can be obtained, especially for compounds of relatively low polarity. If using a small sample plug of 5 μL,¹⁸ the analyte concentration decreased significantly during the SFME, which quenched the chemical exchange with the equilibrium reached at lower concentrations in both phases.

To investigate the impact by the sample volume, SFME was performed to extract the drug verapamil (logP = 3.8) from two urine samples (5 μL and 2000 μL) for a comparison. Ethyl acetate of 5 μL was utilized as the extraction solvent. Ten extraction cycles were applied in each case and the extracts were analyzed by nanoESI MS. As shown in Figure 1b, significantly higher signal-to-noise ratio (S/N) were obtained for 2000 μL urine sample, as indicated by the intensities of the characteristic fragment peaks at m/z 165 and 303 for verapamil. Evidently, an efficient preconcentrating of the analyte from a large sample volume into 5 μL extraction solvent was achieved.

As discussed above, quantitation retaining mandatory precision is an important objective in the development of direct MS analysis for biomedical applications. The use of IS is a standard and effective means for assuring the quantitation precision in MS analysis. Previously, we demonstrated that the incorporation of IS into samples of small volumes can be possibly done in simple means such as pre-coating IS in capillaries for sample transfer,¹⁷ preprinting IS on the paper substrates for paper spray,²¹ and pre-spiking IS into extraction solvent for SFME.¹⁸ Stable isotope labeled compounds are typically used as the IS for MS analysis, since identical chemical properties are preferable for obtaining similar efficiencies in sample treatment and ionization processes. The concentration of the IS, the volume of the sample, and the volumes of the solvents used in the sample preparation, all need to be precisely controlled. For the sampling using SFME discussed here, however, we explored a method to eliminate the need of accurate measurement or control of the volumes of the sample or the extraction solvent, in order to simplify analytical operation for biomedical applications.

The feasibility of this concept can be explained through a study of the chemical exchange process following the derivations below. When an IS is added to the extraction solvent, it diffuses into the sample during the SFME process. The intensity ratio of the analyte (a) and the IS with SFME and nanoESI-MS can be calculated as:

$$\frac{I_a}{I_{IS}}=\frac{k_a}{k_{IS}}\cdot \frac{\frac{1V_S}{D_{IS}{V}_e}+1}{\frac{1}{D_a}+\frac{V_e}{V_S}}\cdot \frac{C_{a-S}^O}{C_{IS-e}^O}$$ Equation 1

Where k_a and k_IS are the overall response constants, $C_{a-s}^o$ and $C_{IS-e}^o$ are the original concentrations of the analyte and the IS, V_s and V_e are the volumes of the sample and extraction solvent, respectively. D_a and D_IS can be estimated by the partitioning factor of the analyte (P_a) and IS (P_IS).

In the case where i) the analyte is relatively polar (small P_a and D_a), ii) the IS is relatively nonpolar (large P_IS and D_IS), and iii) the sample volume V_s is significantly larger than the extraction solvent volume V_e, equation 1 can be simplified as:

$$\frac{I_a}{I_{IS}}=\frac{k_a}{k_{IS}}\cdot D_a\cdot C_{a-s}^o$$ Equation 2

A striking fact indicated by this derivation is that the volumes of the sample and extraction solvent have minimal impact on the calculation of the analyte concentrations. Stable isotope labeled compounds, as simulants to analytes, could not be used as the IS in this case since the conditions i and ii could not be satisfied. It should be noted that k_a/k_IS and D_a should remain nearly constant for all samples. k_a/k_IS could be affected by the matrix components and D_a could be affected by the variation in osmolality.²²

A 3-dimension schematic was drawn according to Equation 1. To simplify the equation, D_a value was set as 1 (a hydrophilic analyte). The change of intensity ratios I_a/I_IS along with the change of partitioning factor ratios D_a/D_IS and volume ratios V_S/V_e is shown in Figure 2. Just as predicted above, when the partitioning factor of the IS was much larger than that of the analyte (e.g. 1000 times) and the volume of sample solution was much larger than that of extraction solvent (e.g. 50 times), the intensity ratio of analyte/IS remained unchanged. It indicated that small changes in the volumes of sample or extraction solvent would not change the actual analyte/IS intensity ratio, which can be used for quantitation by MS.

Figure 2.

Three-dimension schematic showing the analyte/IS intensity ratio predicted for measurements as functions of the analyte/IS partitioning factor ratio and the sample/extraction solvent volume ratio.

As a validation of the predicted quantitation performance by this approach, morphine, a primary metabolite of heroin and a polar analyte (logP = 0.87, condition i), was spiked into synthetic urine for quantitative analysis. Verapamil (logP = 3.8, condition ii), was used as the IS. Relatively consistent intensity ratios were obtained (Figure 3a), regardless the significant variations in the volumes of the sample and the extraction solvent. A noticeable improvement in precision (RSD = 9.6%) was observed when the volume of the sample was increased to 5 mL sample. Similar results were also observed when analyzing cotinine (logP = 0.07) in urine samples, with amitraz (logP = 5) as IS (Figure S2). Better reproducibility in larger volumes is due to a better fit to the condition for Equation 2.

Figure 3.

a) SFME bulk sampling for detection morphine (logP = 0.87) at 500 ng/mL in urine samples of 0.5, 2 and 5 mL, extraction solvent ethyl acetate containing 10 ng/mL verapamil (logP = 3.8). Product ion intensity ratio of morphine (m/z 201.1) and verapamil (m/z 165.1) were recorded. b) Calibration curve for quantitative analysis of cotinine in urine (50–400 ng/mL) using SFME bulk sampling. Ethyl acetate with verapamil 10 ng/mL as IS was used as extraction solvent.

Regardless of the difference of sample volumes, consistent quantitative results (ratios of 0.56–0.60 for morphine) were observed. This can be attributed to the minimal changes in the analyte concentration in sample and the IS concentration in extraction solvent during the SFME process, where the conditions discussed above were met. Since the SFME usually uses an extraction solvent of 5–10 μL, the range of sample volumes should be larger than 0.5 mL. The impacts due to the analyte diffusion into the extraction phase or IS diffusion into the sample phase are negligible, in terms of reducing the original concentrations. Comparing with solid-phase microextraction which has been demonstrated be free of measuring sample volumes,²³ the SFME based method can be more widely applicable because well-developed liquid-liquid extraction methods can be introduced into the system.

The results discussed above shows that a simplified procedure without volume controls could be applicable for biofluid analysis when sample/IS pairs of significantly different partitioning coefficients were used. This is quite different from, or even contradictory to the traditional strategy used for quantitative mass spectrometry analysis, where isotope-labeled internal standards are used to minimize the impact by matrix effects on the quantitation. The robustness of the current method against the variations in samples was testified by analyzing morphine in different lots of plasma samples (N = 4, from different individuals) (Figure S3), and analyzing morphine solutions with different levels of salt contents (50–200 mM) (Figure S4).

In another demonstration, this method was applied to quantitation of cotinine in urine samples, a calibration curve was established in the range of 50 ng/mL to 400 ng/mL. The volume of each sample was picked between 1–4 mL (Table S1). Verapamil (logP = 3.8) was spiked as the IS into the extraction solvent ethyl acetate at 5 ng/mL. The volume of the extraction solvent used for SFME was 10 μL. The MS/MS transitions, m/z 177.2 80.2 and m/z 455.0 165.1, were monitored for cotinine (the analyte) and verapamil (IS), respectively. The I_a/I_IS ratios are plotted as a function of concentration as shown in Figure 3b. With such a large variation in the sample volume (1–4 mL), relatively good linearity (R² = 0.9912) and RSDs (<12%) were still obtained over the cotinine concentration range 50–400 ng/mL.

This method was also demonstrated for monitoring blood drug concentrations, which is important for pharmacokinetic studies. It was applied to quantify one of hydrophilic drugs (some common classes of hydrophilic drugs and metabolites could be analyzed by direct sampling and quantitation are also listed in Table S2), lincomycin (logP = 0.56) in bovine whole blood in a range from 1 ng/mL to 1000 ng/mL. Using glass capillaries with i.d. of 0.50 mm, it was convenient to perform both SFME and nanoESI with extraction solvent of 1–2 μL. The volume of each bovine blood sample was about 100 μL. Amitriptyline (logP = 4.92) was spiked as the IS into the extraction solvent ethyl acetate at 100 ng/mL. The MS/MS transitions, m/z 407.1 126.1 and m/z 278.1 233.1, were monitored for the analyte (lincomycin) and IS (amitriptyline), respectively. The I_a/I_IS ratios were plotted as a function of concentration, as shown in Figure 4. Good linearity with R² of 0.9905 was obtained over the lincomycin concentration range 10–1000 ng/mL, indicating good quantitation performances of this method in drug monitoring in blood samples.

Figure 4.

Calibration curve for analysis of lincomycin in blood (10–1000 ng/mL) using SFME bulk sampling. Ethyl acetate with amitriptyline 100 ng/mL as IS was used as extraction solvent.

In conclusion, a simple quantitation method has been developed for analyzing biofluid samples using mass spectrometry. Efficient extraction of analyte into the organic extraction solvent was achieved. By using nonpolar compounds as the IS for quantitation of polar compounds, a simple analysis protocol was developed without requiring careful measuring or controlling the volume of the sample or the extraction solvent. This represents a huge convenience for on-site chemical analysis, such as the clinical analysis or in-field analysis. As a direct sampling MS method, its quantitation reproducibility may be affected by varied biological matrices and conditions, this can be avoided in targeted analysis for clinical applications where analyses are usually performed on large number of samples with same matrices. By using polar extraction solvents, this method could be modified for the analysis of compounds in non-polar samples such as vegetable oils.