Quantitation of Urinary Acylcarnitines by DMS-MS/MS Uncovers the Effects of Total Body Irradiation in Cancer Patients

Abstract

Acylcarnitines have been identified in human and animal metabolomic-profiling studies as urinary markers of radiation exposure, a result which is consistent with their cytoprotective effects and roles in energy metabolism. In the present work, a rapid method for quantitation of the more abundant acylcarnitines in human urine is developed using a valuable set of samples from cancer patients who received total body irradiation (TBI) at Memorial Sloan Kettering Cancer Center. The method uses solid phase extraction (SPE) processing followed by differential mobility spectrometry (DMS with ethyl acetate modifier) tandem mass spectrometry (ESI-DMS-MS/MS) with deuterated internal standards. The analyzed human urine samples were collected from 38 individual patients at three timepoints over 24 hours during and after the course of radiation treatment, a design allowing each patient to act as their own control and creatinine normalization. Creatinine-normalized concentrations for nine urinary acylcarnitine (acyl-CN) species are reported. Six acyl-CN species were reduced at the 6-hour point. Acetylcarnitine (2:0-CN) and valerylcarnitine (5:0-CN) showed recovery at 24 hours, but none of the other acyl-CN species showed recovery at that point. Levels of three acyl-CN species were not significantly altered by radiation. This rapid quantitative method for clinical samples covers the short and medium chain acylcarnitines and has the flexibility to be expanded to cover additional radiation-linked metabolites. The human data presented here indicates the utility of the current approach as a rapid, quantitative technique with potential applications by the medical community, by space research laboratories concerned with radiation exposure, and by disaster response groups.

INTRODUCTION:

Acylcarnitines are small molecule lipid derivatives comprised of a fatty acid with chain lengths ranging from C2:0-C26:0 linked to carnitine by an ester bond. They have long been identified as markers of inborn errors of metabolism[1]. In addition, altered plasma acylcarnitine profiles have been correlated with type 2 diabetes and obesity[2], and more recently, acetylcarnitine has been identified as a marker indicative of radiation exposure[3–11]. Hence, the quantitation of acylcarnitines has become essential for applications ranging from drug discovery to diagnostic testing.

Metabolomics is an area of study which identifies and quantifies small molecule metabolites in order to understand their characteristic response to genetic or environmental perturbations[12], with applications to biomarker development and pathway analysis. Mass spectrometry (MS) is commonly used for metabolomics applications such as acylcarnitine profiling. Historically, acylcarnitine analysis has been performed with a separation method coupled to MS, such as GC-MS or LC-MS/MS. The existing methods are not suitable for patient triage, or high-volume analysis, as indicated by the following examples of current literature. Meierhofer [13] has recently reported a LC-MSMS analysis covering a broad range of acylcarnitines in fluids and tissues, but acylcarnitines were first derivatized by 3-nitrophenylhydrazine (3NPH), and the required LC run time in the low-resolution system is 45 min. Vernez, et al., described an LC-ESI-MS/MS method for the analysis of urinary acylcarnitines, with LLOQ’s ranging from 0.75–5 μM, depending on the acylcarnitine being analyzed.; however, they use an ion pairing agent in their LC mobile phase, which generally leads to significant ion suppression issues, and their run time is 17 minutes, with an additional 10 minute system re-equilibration time[14]. Peng, et al., reported on an HPLC-MS/MS method for the analysis of human plasma acylcarnitines, with coefficients of variation (%CV) less than 15%; however, they also used an ion pairing agent, heptafluorobutyric acid (HFBA), and their run time was 25 minutes[15]. Minkler, et al., published a 21 minute UHPLC-MS/MS method for the analysis of urinary C5-acylcarnitine isomers, but their sample preparation protocol required derivatization with diisopropylethylamine and fluorphenacyl-trifluoromethanesulfonate, and their LC mobile phase contained the ion pairing agent triethylamine[16]. Sarker, et al, and others have presented FIA-ESI-MS/MS methods for the analysis of acylcarnitines from dried blood spots, with %CV’s within 20%; however, their sample preparation protocols have required kits, which are a convenient yet costly substitution for sample preparation, and their FIA-ESI-MS/MS methods have lacked the selectivity and specificity afforded by a separations technique coupled to the front end of a mass spectrometer[17, 18]. Lastly, Willacey et al. recently reported on a metabolomics method for the analysis of central carbon metabolites, such as acylcarnitines, by LC-MS/MS; however, their method required derivatization with dimethylaminophenacyl bromide (DmPABr) and their run time was 21 minutes[19].

The traditional techniques cited above have quite lengthy run times, and as a result, cannot be utilized for high throughput applications. Consequently, an alternative separation strategy with faster run times and possibly lower, but appropriate, levels of sensitivity and selectivity is preferred. Differential mobility spectrometry (DMS), also known as high field asymmetric waveform ion mobility spectrometry (FAIMS) [20], offers a vapor-phase approach to separation. DMS separations are based on rapidly-switched low/high-field clustering/declustering of analyte ions with gas-phase neutral molecules [21, 22] which results in a difference between high and low field ion mobility. DMS ion filtration occurs in a few milliseconds and provides selectivity and chemical noise suppression[5, 23–28]. Following ionization within the MS source, ions pass through the DMS cell, which is made up of 2 planar electrode plates. An asymmetric electric field waveform, also known as the separation voltage (SV), is applied across the DMS cell. As a result, ions begin to oscillate and may be neutralized on the electrode walls, but specific ions can be guided to reach the MS detector by the application of a quasi-static compensation voltage, COV[29]. The fast DMS ion filtration step suppresses interferents and chemical noise and increases peak capacity[27, 30, 31].

In this article, we describe a high-throughput and quantitative DMS-MS/MS method applied to the analysis of clinical samples collected from cancer patients who had undergone total-body radiation therapy. Possible high-volume or time-limited applications are fast screening for unexpected radiation exposure, to radiation biology studies, and, to space-flight research (NASA, https://www.nasa.gov/hrp/elements/radiation/risks). We eliminated the need for derivatization in our method, relying on a more efficient sample preparation based on batch solid-phase extraction (SPE). In addition, we did not utilize ion pairing agents and used the lowest flow rates compatible with the instrumentation.

In brief, a group of nine short-chain and medium-chain urinary acylcarnitines in urine are analyzed simultaneously, in a single run of less than 1 minute, their absolute concentrations found, and ratios to creatinine reported after a urinary creatinine analysis by a similar method. To the best of the authors’ knowledge, this is the first report of the absolute quantitation of human urinary acylcarnitines by DMS-MS/MS.

METHODS:

Chemicals

Glacial acetic acid and HPLC grade water were purchased from Fisher Scientific (Agawam, MA, USA). Optima LC-MS grade acetonitrile, methanol, ethyl acetate and formic acid were also purchased from Fisher Scientific. The following unlabeled standards were available and purchased from Sigma (St. Louis, MO, USA): acetylcarnitine (C2:0-CN), propionylcarnitine (C3:0-CN), butyrylcarnitine (C4:0-CN), valerylcarnitine (C5:0-CN), hexanoylcarnitine (C6:0-CN), octanylcarnitine (C8:0-CN) and decanoylcarnitine (C10:0-CN). As shown in Table S1 in the supplementary materials, the following stable isotope standards were available and purchased from Sigma (St. Louis, MO, USA), as well: acetyl-d₃-L-carnitine hydrochloride (d₃-C2:0-CN), butyryl-L-carnitine-N-methyl-d₃ (d₃-C4:0-CN), octanoyl-L-carnitine-N-methyl-d₃ (d₃-C8:0-CN), and decanoyl-L-carnitine-N-methyl-d₃ (d₃-C10:0-CN).

Clinical Samples

Georgetown Medical Center received samples from Memorial Sloan-Kettering Cancer Center (MSKCC). After the profiling study [11], samples were transferred to Northeastern University for development of the faster ion-mobility-enhanced quantitation of a broader range of acylcarnitines. Patients had been previously diagnosed with various forms of cancer and had completed chemotherapy. Cancer types included acute myelogenous leukemia, acute lymphocytic leukemia, chronic myelogenous leukemia, Hodgkin’s lymphoma, non-Hodgkin’s lymphoma and myelodysplastic syndrome. Patients of this type undergo an aggregate exposure of total body irradiation (TBI) ranging from 1.25–3.75 Gy prior to hematopoietic stem cell transplantation in either one fraction or three fractions. We selected samples from 38 patients for which all three time points were available (114 samples). These patients received 3 fractions of 1.25 Gy. Although gender differences have been identified [11], we did not perform separate M/F analyses but chose to use a single larger population of 38 patients. This mixed M/F group was treated in the following approximate sequence of collections and radiation exposures, as shown for a smaller group of samples in supplementary data table S1 of the profiling metabolomics study [11]. A schematic approximation of the timing of collections and radiation treatments is shown here.

1. Day 1.

   1. +0 hr: Collection A (approximately 5:00 AM, fasting state). This is the control sample.
   2. +3 hr: Irradiation 1, 1.25 Gy.
   3. +9 hr: Collection B (not fasting). This is the “6 hour” time point.
   4. +10 hr: Irradiation 2, 1.25 Gy.
   5. +15 hr: Irradiation 3, 1.25 Gy.

2. Day 2 +0hr, (Day 1 + 24 hr): Collection C. (fasting state) This is the “24 hour” time point.

These times were subject to variation. Following notation used in the earlier publication [11], we refer to collections A, B, and C as the pre-exposure, 6 hour, and 24 hour time points. It is important to note that diet was not controlled for during this study.

Sample Preparation for Acylcarnitine Analysis of Human Urine

The acylcarnitine internal standard mix was made up of the following four deuterated acylcarnitines, each at a concentration of 200 μM: d3-acetylcarnitine, d3-butyrylcarnitine, d3-octanoylcarnitine, and d3-decanoylcarnitine. Ten μL of the acylcarnitine internal standard mix was spiked into 190 μL of human urine. Following the previous protocol, urine was dried down under nitrogen (N₂) at 60°C and resuspended in 200 μL of methanol. Samples were vortexed and centrifuged at 14,000 rpm for 5 min. at room temperature. Urinary acylcarnitines were selectively isolated by solid-phase extraction (SPE) using SEP-PAK VAC 6 cc silica gel cartridges (Waters, Milford, MA, USA) following a literature procedure. This cartridge type is useful for acylcarnitines because it adsorbs analytes of even weak hydrophobicity from aqueous solution. Acylcarnitines were eluted off the SPE cartridges with 1.5 mL methanol:water:acetic acid (50:45:5, v/v/v). Extracts were then dried down under N₂ and resuspended in 1 mL of methanol:water:formic acid (70:30:0.1, v/v/v). Samples were vortexed and centrifuged at 14,000 rpm for 5 min. at room temperature and ultimately transferred to LC-MS vials (Agilent, Palo Alto, CA, USA). Flow injection analysis (FIA) was utilized for sample introduction into the ESI-DMS-MS/MS platform.

Sample Preparation for Urinary Creatinine Analysis

A dilution method was used to measure creatinine, due to the normal high urinary levels. Five μL of urine were added to 995 μL of the creatinine extraction solvent, acetonitrile:water:formic acid (50:50:0.1, v/v/v), and vortexed. Ten μL of this solution was further diluted into 140 μL of methanol:water:formic acid (70:30:0.1, v/v/v) and vortexed. One-hundred μL of this solution was then added to 100 μL of 10 μM deuterated creatinine in the same solvent, resulting in a 6000-fold dilution. Samples were vortexed and centrifuged at 14,000 rpm at room temperature for 5 min and transferred to LC-MS vials.

Flow Injection Analysis

An Acquity UPLC, Sample Manager and Sample Organizer (Waters, Milford, MA, USA) were utilized for analysis. All Acquity components were controlled through companion software installed within Analyst 1.6.2 (Sciex, Framingham, MA, USA). The FIA mobile phase was methanol:water:formic acid (70:30:0.1, v/v/v) and its flow rate was 50 μL/min, the minimum flow usable for FIA with this UPLC, selected because low flow rates minimize competitive ionization. The strong wash was water and the weak wash was acetonitrile:water (1:1, v/v). Sample injection volumes were 10 μL.

To prevent carryover between samples, and monitor the instrument performance, blanks and quality-control standards were inserted into the scheduling of sample analysis. These were monitored during the analysis, both for carryover and for maintenance of sensitivity.

Differential Mobility Spectrometry

The SelexIon (Sciex, Framingham, MA, USA, https://sciex.com/technology/selexion-technology) was utilized for all DMS analyses. The “low” setting was selected for the DMS temperature, which equates to 150°C. Ethyl acetate was the DMS modifier (MOD) and its composition was set to “low”, which corresponded to 1.5% v/v. The DMS offset was −3.0V and the DMS resolution enhancement (DR) was turned off. The separation voltage (SV) was 3500 V. The compensation voltage (COV) for each acylcarnitine was tuned prior to sample analysis. All DMS data were acquired with Analyst 1.6.2 (Sciex, Framingham, MA, USA).

Mass Spectrometry

MS and MS/MS data were acquired with a QTRAP 5500 mass spectrometer with SelexIon and Analyst 1.6.2 (Sciex, Framingham, MA, USA). Unlabeled acylcarnitines and stable isotope acylcarnitines were analyzed with positive electrospray ionization (ESI+) and quantified in the multiple reaction monitoring (MRM) mode (Supplementary Table S2). All source conditions were the same for each acylcarnitine. The collision gas (CAD) was set to medium and the curtain gas (CUR) was 20 psi. The ion source gas 1 (GS1) pressure and the ion source gas 2 (GS2) pressure were both set to 60 psi. The source temperature (TEM) was 600°C and the ion spray (IS) voltage was 4500 V. Declustering potential (DP) was 60 V and the entrance potential (EP) was 10 V. The collision energy (CE) was 21 V and the collision cell exit potential (CXP) was 7 V. Even though the product ions are the same for each acylcarnitine, we elected to use an MRM method, as opposed to a precursor ion scan method, because each acylcarnitine has a unique optimal DMS compensation voltage, as was discussed in the previous DMS section. Unlabeled creatinine and deuterated creatinine were also analyzed with ESI+ and quantified in MRM mode. FIA peak integration was performed with MultiQuant software 2.1 (Sciex, Framingham, MA, USA). Data reduction, data visualization and statistical analyses were performed with Microsoft Excel and GraphPad Prism 8.02.

SPE Efficiency and ESI Matrix Effects

The effectiveness of SPE for sample concentration and reduction of matrix effects is discussed by Mallet, et al. [32], and by Dams, et al. [33], as well as in the reference recommending our protocol which also reports separation efficiency [34]. The first two papers evaluate SPE in comparison to cleanup by protein precipitation, by dilution, and to direct flow injection. SPE is especially effective when it is selected for a particular targeted analysis, as in the current case, and in earlier work from this laboratory[35].

Ion suppression in ESI by matrix effects is described in Panuwet et al.[36] as occurring when the matrix (1) blocks analyte from the droplet surface, (2) prevents charging of the analyte in the droplet, or (3) steals charge in the gas phase. These matrix effects are all reduced at lower flow rates because of smaller droplet size and greater available charge per molecule, as discussed by K. Tang, et al.[37]. We use the minimum practicable flow rate with the current instrumentation (50 μL/min., about 100X lower than typical LC flows), and a high desolvation temperature for quick droplet dispersal. In the creatinine case, high dilution in solvent minimizes matrix effects (see Stahnke, et al.[38], as well as Dams, et al. [33]), by preventing analyte-matrix interactions, and increasing the available droplet charge per analyte molecule.

DMS-MS Method Performance

DMS parameters such as separation voltage (SV) and compensation voltage (COV) were optimized for acylcarnitines during the method development phase. Ethyl acetate was found to be the best performing modifier for COV shift and chemical noise suppression, with the modifier level (MOD) set to low, corresponding to 1.5% (v/v) in the curtain gas. Prior to sample analysis, COV values were readjusted once using a standards mixture, taking about 20 min. Fig. 1 presents a 3D plot of the DMS separation of acylcarnitines based on experimentally obtained COV and m/z values, including observed DMS peak widths (Supplementary Table S3). It is important to note that the COV values are not affected from the zwitterionic nature of the acylcarnitines because they are resuspended in 70% methanol, thereby resulting in protonation of the carboxylate group and a positively charged center. Thus, DMS provides enhanced resolving power along with a substantial reduction in matrix-based chemical noise, as has been described in numerous publications [5, 23–25].

Fig. 1. Three-dimensional separation of acylcarnitine standards by DMS-MS, under conditions described in the experimental section based on experimental observations of DMS peak widths.

Ethyl acetate at the low setting was used as the modifier (MOD). The size of the bubble indicates the COV peak width at maximum height

RESULTS AND DISCUSSION:

Pilot Study Identifying Responding Acylcarnitines

A reduced-size pilot study prior to the full analysis allowed us to select acyl carnitines with observable concentration levels, and apparent radiation sensitivity. A simple assessment of relative levels of thirteen potential urinary acylcarnitines (C2 through C16, see supplemental table S2) was performed. Relative intensities were determined using one deuterated acylcarnitine internal standard for each chain-length group, the short, medium, and long chain acycarnitines. The three added internal standards were deuterated acetylcarnitine (C2:0-CN-d3); deuterated decanoylcarnitine (C10:0-CN-d3); and deuterated palmitoylcarnitine (C16:0-CN-d3). Nine patient urine samples were analyzed. They were randomly chosen, three samples from each time point. The pilot study results appear in Fig. 2, which is divided into two panels, with the more abundant species in Figure 2a and the less abundant in Figure 2b. Acetylcarnitine (C2:0-CN) is the most abundant urinary acylcarnitine, followed by butyrylcarnitine (C4:0-CN) and valerylcarnitine (C5:0-CN). Since valerylcarnitine (C5:0-CN) is an odd chain acylcarnitine, its identification and relative abundance were somewhat unexpected, yet quite noteworthy, and supported the clear need to perform a comprehensive investigation of all clinical samples. In addition, octenoylcarnitine (C8:1-CN), from the mostly-unexplored monounsaturated acylcarnitines, was also significant. A reduction at 6 hours after irradiation was detected for almost every detected acylcarnitine, followed by recovery at 24 hours in some cases.

Fig. 2. Pilot study performed with small subset of human urine samples (n=3).

(a) Most abundant human urinary acylcarnitine species. (b) Least abundant human urinary acylcarnitine species. Error bars represent the standard error of the mean (SEM) for each biological replicate

For the next stage of the analysis, nine acycarnitines were selected (see supplementary table S1). The selected compounds were the short and medium chain saturated acylcarnitines, of lengths C2:0-CN through C9:0-CN, and the mono-unsaturated C8:1-CN. The long-chain acylcarnitines were not present at high enough levels for accurate quantitation by this method.

Acylcarnitine Calibration Curve Preparation in Rat Urine Matrix

Prior to the analysis of human urine samples, the DMS-MS/MS method was validated in order to assess the linear dynamic range for the selected acylcarnitine species. Due to the limited quantity of human urine from the clinical study, Sprague Dawley (SD) rat urine matrix (provided by Pfizer) was used to generate calibration curves for the purpose of validation. Structures S1-S10, found in the supplementary materials, illustrate each acylcarnitine that was used for calibration curve preparation (taken from www.lipidmaps.org). Stable isotope acylcarnitine internal standards were not available for each acylcarnitine. Compounds were calibrated using deuterated standards of similar chain length (see supplemental table S1). As a result, only four deuterated acylcarnitine internal standards (IS) were used in the preparation of the calibration curves: deuterated acetylcarnitine (C2:0-CN-d3); deuterated octanoylcarnitine (C8:0-CN-d3); deuterated decanoylcarnitine (C10:0-CN-d3); and deuterated palmitoylcarnitine (C16:0-CN-d3. In total, three calibration curves were prepared, with two of the curves containing three different acylcarnitine species each, along with the third curve which contained four separate acylcarnitine species. Each calibration curve was run in triplicate. The linear dynamic range spanned from 2–3 orders of magnitude, regardless of the acyl chain length of the internal standard utilized. In addition, the R² values ranged from 0.9948 to 0.9998 for each acylcarnitine calibration curve. Hence, the calibration curve data provided us with confidence in using species of different chain length for the quantitation of acylcarnitines that were not available as standards.

Validation by Blind Urine Analyses

The DMS-MS/MS method was evaluated through blind analyses of five acylcarnitine species at three different concentrations spiked into rat urine matrix. Eleven rat urine samples were spiked with unlabeled acylcarnitine standards at varying concentrations by one analyst and then analyzed by a different analyst. Samples 1–10 each contained only 1 unlabeled acylcarnitine standard, whereas sample 11 was spiked with a mixture containing all 5 unlabeled acylcarnitine. Fig. S1 and Table S4 (supplementary materials) compare the experimental concentrations to the theoretical concentrations, showing an adequate level of agreement in the relevant biological ranges above 6 μM. These data indicate that acylcarnitine standards with different chain lengths could be used for the quantitation of acylcarnitines, allowing us to proceed to the analysis of the human samples. Lastly, it is important to note that only relative concentrations should be reported for urinary acylcarnitine species that fall below 6 μM when using this method for analysis.

Randomized Analysis of All Human Urine Samples

The following targeted acylcarnitine species were unavailable as deuterated standards: heptanoylcarnitine (C7:0-CN), octenoylcarnitine (C8:1-CN), and nonanoylcarnitine (C9:0-CN). Consequently, these particular acylcarnitines were quantified to species with similar chain lengths, an approach that is used in most other reported acylcarnitine analysis methods (for example, Meierhofer [13] ). Four stable isotope internal standards were found to be adequate for the quantitation of the nine human urinary acylcarnitines listed in Table S1, namely, the saturated acylcarnitines of chain lengths 2 through 9 and monounsaturated C8:1-CN. Meierhofer has made use of an extended analysis based on multiple fragment ion detection, consistent LC elution times in a very long chromatographic analysis, and high resolution Orbitrap spectra to disentangle structural isomers. Our identification of the targeted compounds is based on DMS COV parameters fitting correctly in the expected trend, and the dominant parent / fragment m/z values. As a result, contributions of multiple isomeric acylcarnitines are not completely excluded in this method. Nonetheless, the correlation with radiation exposure at this level of resolution is seen at a significant level.

A pooled human urine sample broadly characteristic of the entire cohort was used for calibration curve generation and for the patient sample analysis, prepared as follows. Initially, the 114 frozen clinical urine samples were globally randomized into 4 batches, to be processed over 3.5 days. On day 1 of processing, the 32 samples from batch 1 were thawed and arbitrarily chosen for the generation of a pooled human urine sample, which was then used for calibration curve generation. It is important to note that the human urine samples chosen for the pooled samples were made up of various subjects at various timepoints, thus providing a truly heterogeneous pooled sample.

The pooled human urine was used as the matrix for the preparation of calibration curves for the seven saturated normal-species acylcarnitines which were available (C2:0-CN, C3:0-CN, C4:0-CN, C5:0-CN, C6:0-CN, C8:0-CN, and C10:0-CN) (Fig. S2 in supplementary materials), thereby allowing for the absolute quantitation of urinary acylcarnitines. The four available short and medium-chain stable-isotope internal standards (d3-Cn:0-CN, n = 2, 4, 8, 10) were used for the calibration and quantitation of the nine selected human urinary acylcarnitine species (supplementary table S1, also including calibration curve slopes). To compensate for endogenous levels in the matrix, we follow the guidance of Mani, Abatiello, and Carr [39], to interpret calibration curve intercepts as measures of endogenous levels, and use the slopes only to calculate total concentrations. Creatinine was measured in separate runs of the 6000-fold solvent-diluted form of each of the clinical samples, using a neat creatinine calibration curve.

Acylcarnitine concentrations were normalized to their corresponding creatinine concentrations as shown in Fig. 3, with significance levels. Since urine samples were collected at multiple time points including pre-irradiation, each patient provides their own control level and has individual time-resolved creatinine normalization, so it is appropriate to apply repeated-measures statistical analysis. Statistical significance of the clinical data set was determined by repeatable measures one-way ANOVA performed in GraphPad Prism 8.02. The normalized concentrations of most acylcarnitine species showed a reduction at the 6-hour time point. Acetylcarnitine (C2:0-CN) and valerylcarnitine (C5:0-CN) had recovered at Day 2, the 24-hour time point (Fig. 3a and 3b). Butyrylcarnitine (C4:0-CN), heptanoylcarnitine (C7:0-CN) and octenoylcarnitine (C8:1-CN) levels were reduced at 6 hours, but had not recovered at 24 hour (Fig. 3c, 3d and 3e). Nonanoylcarnitine (C9:0-CN) was significantly reduced at both 6 hours and 24 hours post irradiation (Fig. 3f). Lastly, propionylcarnitine (C3:0-CN), hexanoylcarnitine (C6:0-CN) and octanoylcarnitine (C8:0-CN) did not significantly change as a result of radiation (Fig. 3g, 3h and 3i).

Fig. 3. FIA-DMS-MS/MS analysis of urinary acylcarnitines from 38 cancer patients who underwent radiation therapy.

Patient acyl-CN concentrations were normalized to their corresponding creatinine concentrations. Acyl-CN species are grouped based on response to radiation and are presented accordingly (a-i). Statistical significance was determined by repeated measures one-way ANOVA. Error bars represent the standard error of the mean (SEM) of the entire cohort for each time point. (n.s = not significant)

Between-patient Variability

As is the case with most clinical studies, considerable variability was observed within this group of cancer patients. Individual patients had different forms of cancer, although all suffered from a hematopoietic malignancy and received a variety of treatment regimens prior to TBI. These was a disease-specific drug therapy regimen, unique to each patient’s case. Lastly, age, gender and diet were not controlled. All of these factors, along with the inherent genetic variability found in human populations, led to patient to patient variability. We have seen that the repeated-measures analysis of variance identifies the overall unity of response for the typical patient shown by the patient repeated-measures p-values in Figure 2. Using a statistical approach that does not include the fact that each patient was tracked in time exposes patient-level variability within this clinical study.

In order to demonstrate the variability amongst the cancer patients, box and whisker plots were generated for each acylcarnitine using all patient data. The bottom and top of each box signify the 25^th and 75^th percentile of the data, while the line within the box indicates the data median. The whiskers extend to the upper and lower extreme data points, thereby displaying the variability beyond the upper and below the lower quartiles. It is interesting to note the differences in variability that occurred based on time point. For example, acetylcarnitine (C2:0-CN) and valerylcarnitine (C5:0-CN) exhibit the most variability 24 hrs. after irradiation, yet minimal variability at the earlier time points (Fig. 4 a-b). However, hexanoylcarnitine (C6:0-CN) and octanoylcarnitine (C8:0-CN) display the greatest variability 6 hrs. after irradiation (Fig. 4 g-h). Butyrylcarnitine (C4:0-CN), heptanoylcarnitine (C7:0-CN), octenoylcarnitine (C8:1-CN) and nonanoylcarnitine (C9:0-CN) all had increased variability at the pre-irradiation time point (Fig. 4 c-e, i). Lastly, the variability for propionylcarnitine (C3:0-CN) which Fig. 3g shows was unaffected by the radiation, was comparable for each of the three time points (Fig. 4 f). Due to the high between-patient variability, and the experimental design with individual controls and creatinine levels, only the repeated-measures statistical analysis shown in Fig 3 was performed.

Fig. 4. Box and whisker plots for each acylcarnitine.

Acylcarnitine species are grouped based on response to radiation and are presented accordingly (a-i). The bottom and top of the box represent the 1^st and 3^rd quartile, while the line inside the box indicates the median. The whiskers represent the lower and upper extreme data points and also demonstrate the variability beyond the lower and upper quartiles

Biological Effect of Radiation Exposure

The data presented above clearly demonstrate that radiation therapy or other radiation exposures affect the metabolic profile, and particularly the urinary acylcarnitine profile. However, in order to truly understand the consequence of radiation exposure, it is important to dive into the biology associated with acylcarnitines. Acylcarnitine chain length is determined by fatty acid or amino acid β-oxidation. Even-chain acylcarnitines become elevated as a result of incomplete fatty acid β-oxidation, as opposed to odd-chain acylcarnitines which are produced during amino acid catabolism[40]. Butyrylcarnitine (C4:0-CN) differs from other acylcarnitine species in that it can be generated from either pathway.

Acetylcarnitine (C2:0-CN) is produced by combining acetyl CoA with carnitine, a reaction catalyzed by carnitine acetyltransferase (CrAT). The main function of CrAT is to synthesize acetylcarnitine, which results in its transport into and out of the mitochondrion. Altered CrAT results in an accumulation of acetyl-CoA, and ultimately, an inhibition of pyruvate dehydrogenase (Fig. 5)[41]. Acetyl-CoA can then enter the tricarboxylic acid (TCA) cycle via pyruvate or fatty acid oxidation. Hence, modulation of CrAT could result in alterations within the TCA cycle, thus affecting fuel production and fuel delivery.

Fig. 5.

Schematic of carnitine pathways and reactions [41] (by permission of Springer Nature).

CrAT also plays a role in various types of cancer. For example, CrAT is elevated in prostate cancer cells[41]. In addition, it may be involved in histone acetylation in cancer cells[41]. Since cell growth can be regulated by acetyl CoA through histone acetylation, it is certainly possible that radiation therapy could be altering CrAT and acetyl CoA levels, thereby leading to changes in urinary acylcarnitine profiles.

CONCLUSIONS:

Radiation therapy is a common approach used for the prevention of cancer cell proliferation; however, the metabolic side effects of this treatment have not been widely studied. In recent years, acylcarnitines have been characterized as potential biomarkers of radiation exposure and mined for biodosimetry purposes [3–10]. We previously reported on the use of SPE ESI-DMS-MS/MS to quantify the long-term (7 day) response of non-human primate (NHP) acetylcarnitine (C2:0-CN) upon exposure to radiation[5], which is increasingly upregulated with dose at that delay. This work has the rare opportunity to quantify the short-term metabolic effects of radiation therapy in humans across a panel of short and medium chain acylcarnitines by further targeted development of that rapid quantitative method.

Differential mobility spectrometry is an analytical technique that separates molecules based on their polarity and conformation. Major advantages of DMS include the reduction of chemical background and the enhancement of signal-to-noise (S/N) and, when coupled to tandem mass spectrometry, provides improved quantitation. FIA-DMS-MS/MS acquisition times are on the order of 30–60 seconds, remarkably faster than LC or GC run times which typically range from 5–10 minutes.

Here we present the data from a clinical study involving 38 cancer patients recruited from MSKCC. Patients underwent radiation treatment and urine was collected at 3 time points (pre-exposure, 6 hours post-exposure, and 24 hours post-exposure). We employed the Sciex SelexIon DMS-MS/MS QTRAP 5500 platform on SPE-preprocessed samples coupled to flow injection analysis (FIA), resulting in fast analysis times of less than 0.5 minutes per injection with no chromatographic separation. Our acylcarnitine method was able to measure a number of acylcarnitine species per injection by synchronized switching of DMS and mass spectrometric settings, observing both the targets and the internal standards. After a pilot study to select the most significant species, nine higher-level urinary species were detected and quantified from the human urine samples. Most acylcarnitine species showed reduced levels in urine at the 6-hour point, excepting acetylcarnitine (C2:0-CN) and valerylcarnitine (C5:0-CN) which recovered to pre-irradiation levels at the 24-hour point. Propionylcarnitine (C3:0-CN), hexanoylcarnitine (C6:0-CN) and octanoylcarnitine (C8:0-CN) were not significantly altered as a result of radiation treatment.

This chromatography-free absolute quantitation of human urinary acylcarnitines from patients undergoing radiation therapy shows that FIA-DMS-MS/MS following batch SPE -preprocessing can be a rapid, quantitative technique applicable to initial radiation exposure screening. Moreover, the data suggest that the cancer cell death caused by radiation exposure could be leading to varied CrAT levels, thus resulting in altered urinary acylcarnitine profiles. The findings and methodology described in this research project may benefit the medical community as a result of the insights made into the underlying biology associated with radiation treatment. In addition, space research laboratories such as NASA may utilize portable DMS instruments to assess radiation exposure during missions in space. Lastly, disaster response groups may adopt DMS as an initial technique for the screening of patients following a radiation release.