Development and validation of a highly-sensitive, quantitative LC-MS/MS assay to evaluate plasma oxytocin

Graphical abstract

Highlights

• •LC-MS/MS method combined with updated SPE for quantification of oxytocin in plasma.
• •Fully validated according to CLSI guidelines.
• •LLOQ of 1 ng/L achieved while maintaining a low degree of measurement uncertainty.

Abstract

Introduction

Oxytocin is a 9-amino acid peptide that serves as neuromodulator in the human central nervous system. This peptide is implicated in the regulation of diverse behaviors and plays a significant role in positive social interaction. Currently, oxytocin levels are measured using immunoassays. However, these methods have several limitations that can lead to false results and erroneous interpretation. Given the remarkably low endogenous level of oxytocin in human plasma (low ng/L levels), we developed and rigorously validated a novel and highly sensitive LC-MS/MS method for oxytocin quantification in plasma.

Methods

Oxytocin was initially extracted using solid-phase extraction with an Oasis HLB 30 mg plate and then subjected to LC-MS/MS analysis. PBS-0.1 % BSA served as surrogate matrix for the preparation of validation samples and the calibration curve, ensuring no endogenous interference. The validation design followed the Clinical Laboratory Standards Institute guidelines. Precision, accuracy, and measurement uncertainty were determined using single-nested analysis of variance and e.noval software.

Results

A lower limit of quantification of 1 ng/L was achieved. The method was validated for oxytocin concentrations ranging from 1 ng/L to 75 ng/L, with precision (coefficient of variation) below 10 %, accuracy ranging from 94 % to 108 %, and measurement uncertainty below 15 %.

Conclusion

In this work, we developed and validated a highly sensitive LC-MS/MS method for the quantification of oxytocin in plasma. Our novel methodology is well-suited for clinical applications.

1. Introduction

First synthesized by du Vigneaud and co-workers in 1953 [1], oxytocin, a 9-amino acid peptide, is primarily produced by magnocellular and parvocellular neurons in the hypothalamus [2]. Once released into the plasma, this peptide has a half-life of a few minutes before being excreted by the kidneys [3]. Oxytocin is present at < 8 ng/L in human plasma [4] and has been studied for its potential behavioral properties, including its impact on learning, memory, and its crucial role in pair bonding [5]. Oxytocin functions in birth, lactation, and parenting, and may enhance pro-social interactions, such as maternal care, trust, partner preference, and social recognition [5]. Additionally, oxytocin plays an essential role in the regulation of stress and anxiety [6], [7]. It can function as a stress-coping molecule, an anti-inflammatory agent, and an antioxidant, with protective effects, especially in the face of adversity or trauma [8].

Early-life stress can have a long-term influence on the oxytocin system by altering the expression and signaling of oxytocin receptors. This can lead to deficits in social behavior, emotional control, and stress responses that may increase the risk of anxiety, depression, and other stress-related neuropsychiatric diseases [9]. Emerging evidence suggests that oxytocin-system dysfunction might play a pleiotropic role in the etiology of psychiatric disorders. Oxytocin and its receptor are extensively altered in depressive disorders, and exogenous oxytocin administration has been shown to alleviate depression by modulating the stress-Hypothalamic-Pituitary-Adrenal (HPA) axis-immune crosstalk [10]. Some studies have shown reduced plasma oxytocin levels in children with autism spectrum disorders (ASDs) [11]. Patients with ASDs [12] often exhibit significant deficits in social and emotional competence, closely associated with an imbalance in the oxytocinergic system [13]. Consequently, measuring human oxytocin in the context of socioemotional state may provide valuable information about individual differences [14].

Although an abundance of scientific literature implicates oxytocin in health and disease, the relationships between endogenous oxytocin and responses to exogenous oxytocin are not easily identifiable [8]. Sources of variation in outcomes include genetics, epigenetics, and early life adversity. Regardless, it is becoming clear that assessment of plasma oxytocin levels under basal and/or stimulated conditions has the potential to reveal individual vulnerabilities in the activity or reactivity of the oxytocinergic system [14]. Hence, there is reasonable expectation that oxytocin may be useful as a remedy for the previously mentioned morbidities.

Recent exploration into nasal administration of oxytocin by several groups aimed to ameliorate symptoms such as asocial behavior [15], [16], [17]. However, discrepancies were noted in experimental design, pharmacological effects, and the type of biological sample used [14]. Several laboratories have developed radio immunoassays (RIA) techniques [18], becoming the gold standard in oxytocin quantitation. However, several criticisms have emerged about this methodology. Indeed, the lack of specificity and sensitivity has impeded research progress due to data that is difficult to interpret and results that are difficult to reproduce. Factors may explain the differences between the results obtained by the various immunoassays include the presence or absence of an extraction step before the actual assay, lack of sensitivity, cross-reactivity, and the presence of heterophilic antibodies [19].

To overcome these challenges, liquid chromatography − mass spectrometry (LC-MS) methods have been developed to more accurately measure the levels of oxytocin in various biological samples, including plasma, serum, urine, saliva and cerebrospinal fluid [20], [21], [22], [23], [24], [25]. Recognized as the gold-standard of analytical methods [14], these techniques offer specificity and selectivity superior to immunoassays. Sensitivity, with a lower limit of quantification (LLOQ) reaching 1 ng/L, has been achieved for oxytocin with high-resolution mass spectrometry (HRMS) [26] and two-dimensional liquid chromatography (2D-LC) [24]. However, these instruments are expensive and are not generally accessible to most laboratories. The existing methods for analysis of oxytocin are summarized in Table 1.

Table 1.

Summary of existing methods for the quantitation of oxytocin in human plasma. SPE: Solid phase extraction, SALLE: Salt assisted liquid–liquid extraction, PPT: Protein precipitation.

Method	Sample preparation	LLOQ	Reference
LC-HRMS	SPE	5 ng/L	[22]
Nano LC-HRMS	SALLE	1 ng/L	[23]
2D LC-MS/MS	SPE	1 ng/L	[24]
Nano LC-MS/MS	PPT	5 ng/L	[25]

Recognizing the need for a reliable and sensitive assay for oxytocin quantitation that is accessible to most laboratories, we aimed to develop a LC-MS/MS method for the accurate measurement of oxytocin in human plasma following the Clinical and Laboratory standards Institute (CLSI) guidelines. Our method focused on achieving high sensitivity to quantify oxytocin at basal levels in plasma while maintaining low measurement uncertainty.

2. Materials and methods

2.1. Chemicals and reagents

Acetonitrile (ACN) and water (H₂O) (LC-MS grade) were purchased from Biosolve (Dieuze, France). Phosphoric acid (H₃PO₄) and trifluoroacetic acid (TFA) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Formic acid (FA) and Phosphate Buffered Saline (PBS) were obtained from ThermoFisher Scientific (Kandel, Germany). Human oxytocin (96 % purity) was purchased from ThermoFisher Scientific (Kandel, Germany) and oxytocin-d5 (Oxytocin-(leucine-5,5,5-d3, glycine-2,2-d2) trifluoroacetate salt) stable isotopically-labeled internal standard (SIL-IS) was obtained from Sigma-Aldrich (St. Louis, MO, USA). Bovin serum albumin (BSA) was obtained from Merck KGaA (Darmstadt, Germany). Remnant EthyleneDiamineTetraacetic Acid (EDTA)-plasma samples were collected for routine analysis purposes. No ethical committee was necessary for this procedure according to the Belgian Law of December 19th, 2008, on remnant human corporal material. Cow plasma was collected in Citrate-Phosphate-Dextrose solution with Adenine (CPDA) blood bag (500 mL) by the Veterinary Faculty of the University of Liège (Belgium). Blood samples were taken from subjects of the experimental farm following a generic agreement given by the Veterinary Faculty of the University of Liège and approved by the farm director according to the established procedures. In agreement with the university hospital-faculty ethics committee of Liège concerning blood samples taken from healthy volunteers as part of the activities of the Clinical Chemistry department, human plasma was also collected. The samples were then centrifuged at 3500 rpm during 15 min at 21 °C, aliquoted in 15 mL falcon tubes or 2 mL Eppendorf vials, and stored at −20 °C.

2.2. Standards, calibration and quality controls

Human oxytocin (5 mg) and oxytocin-d5 as the stable isotopically-labeled internal standard (SIL-IS) (1 mg) were dissolved in H₂O, 5 % ACN, 0.1 % FA to create stock solutions. The solvents were chosen for compatibility with mass spectrometry (MS), with ACN used to maximize the solvation process. The resulting solutions were aliquoted (50 µL) into 1.5 mL Eppendorf vials, centrifuged, and stored at −80 °C. Working and spiking solutions were prepared in H₂O, 5 % ACN 0.1 % FA. All solutions were prepared in Protein LoBind Eppendorf tubes. The purity of the standard solutions, as provided by the manufacturer, was carefully considered to ensure the accuracy of the measurements.

For the calibration curve, 10 spiking solutions were prepared to achieve the following concentrations: 27 ng/L, 54 ng/L, 108 ng/L, 162 ng/L, 270 ng/L, 540 ng/L, 1080 ng/L, 1620 ng/L, 2168 ng/L and 2071 ng/L. PBS with 0.1 % BSA was used as the surrogate matrix, ensuring no other interference at the retention time of oxytocin (see Fig. 1). Subsequently, 1300 µL of surrogate matrix, in 2 mL Protein LoBind Eppendorf tubes, was spiked with 50 µL of calibration solution to obtain tubes with final oxytocin concentrations of 1 ng/L, 2 ng/L, 4 ng/L, 6 ng/L, 10 ng/L, 20 ng/L, 40 ng/L, 60 ng/L 80 ng/L, and 100 ng/L. A blank containing only the SIL-IS and a double blank, without oxytocin or SIL-IS, were always included.

Fig. 1.

Total ion current (TIC) chromatogram of a double blank sample.

For the validation samples, eight distinct concentration levels were prepared in the same way as calibration samples with concentrations of 1 ng/L, 2.5 ng/L, 5 ng/L, 7.5 ng/L, 12.5 ng/L, 25 ng/L, 50 ng/L, and 75 ng/L. These samples were prepared using PBS with 0.1 % BSA (w/w).

Three human validation samples spiked with 5, 50 and 100 ng/L of oxytocin, and a calibration curve were also prepared in human plasma.

All working and spiking solutions, as well as calibrators and validation samples were freshly prepared for each run.

2.3. Sample preparation

Before sample extraction, spiked plasma samples, validation samples, or calibrators were thawed and centrifuged for 5 min at 3500 rpm and 21 °C to ensure the homogeneity. Subsequently, 1350 µL of spiked plasma, validation samples, or calibrators was transferred to a 2 mL Protein LoBind Eppendorf tubes and 500 µL of 5 % H₃PO₄ was added, as advised by Zhang et al [24]. The use of H₃PO₄, a strong acid, served to denature the proteins in plasma and release the oxytocin from protein binding sites, with the goal of increasing extraction recovery. Following this, the samples underwent a 15 min incubation period at room temperature to ensure the effective release of oxytocin. Next, SIL-IS spike solution (50 µL for 500 ng/L final concentration, Fig. 2) was added to the samples, validation samples and calibrators, followed by homogenization using a mini analog vortex mixer (30 sec, high speed). The samples, validation samples and calibrators were finally centrifuged for 1 min at 13000 rpm and 21 °C. Following the methodology of Zhang et al [24], sample clean-up and preconcentration were accomplished utilizing an Oasis HLB 96-well plate with 30 µm particles. Wells were conditioned with 1 mL ACN followed by 1 mL of 1 % H₃PO₄. Subsequently, 1900 µL of diluted samples was loaded and washed sequentially with 1 mL of 1 % H₃PO₄ followed by 1 mL of H₂O: ACN (95:5) with 0.1 % TFA, and 1 mL H₂O. Elution was carried out with 450 µL of ACN: H₂O (60:40). Following dry evaporation (N₂, 45 L/min, 45 °C) using a Turbovap® 96 double evaporator (Biotage, Uppsala, Sweden), 200 µL of ACN: H₂O (90:10) were added to precipitate the residual proteins. The samples were transferred to a second plate and subjected to dry evaporation (N₂, 45 L/min, 45 °C). Finally, the samples were reconstituted in 80 µL of H₂O: ACN (95:5) with 0.1 % FA. The detailed procedure for sample preparation is shown in Fig. 3.

Fig. 2.

Oxytocin SIL-IS chromatogram at a 500 ng/L concentration (after SPE treatment) for the transition 507.5 > 937.2.

Fig. 3.

Sample preparation scheme. The SPE plate was conditioned with 1000 µL of ACN followed by 1000 µL of H₃PO₄ 1 %. 1900 µL of diluted sample was added and washed consecutively by 1000 µL of H₃PO₄ 1 %, 1000 µL of H₂O: ACN (95:5) with 0.1 % TFA and 1000 µL of H₂O. Elution required 450 µL of 60 % ACN. After protein precipitation and evaporation, samples were reconstituted with 80 µL of H₂O: ACN (95:5) with 0.1 % FA.

2.4. LC-MS/MS conditions

Analysis was performed using an ACQUITY UPLC H-Class PLUS system (Waters, Milford, USA) coupled to a XEVO TQ-XS mass spectrometer, a triple quadrupole instrument (QqQ), also from Waters (Milford, USA). Electrospray ionization was employed in positive mode (ESI + ). Chromatographic separation was achieved on an ACQUITY UPLC BEH C18 130 Ǻ (2.1 x 100 mm, 1.7 µm) column from Waters (Milford, USA), maintained at 50 °C. The mobile phases for the chromatographic separation consisted of H₂O (phase A) and ACN (phase B) both containing 0.1 % FA. The flow rate was set at 0.25 mL/min with a percentage of phase B changing as follows: 0 min: 5 %, 1 min: 5 %, 11 min: 90 %, 11.1 min: 95 %, 13 min: 95 %, 13.1: 5 % with an equilibration of 2 min at 5 %. Samples were stored in an autosampler at 10 °C, and the optimum injection volume was determined to be 40 µL.

For MS/MS, two Multi Reaction Monitoring (MRM) transitions were selected and monitored for both oxytocin and SIL-IS. All the MS parameters were optimized to achieve the required sensitivity within the range of 1 ng/L. The specific MRM transitions with the corresponding compound and source parameters are shown in Table 2.

Table 2.

Mass spectrometer parameters: MRM transition, compound and collision energy.

MRM parameters
	Precursor ion (m/z)	Product ion (m/z)	Collision energy (V)	Fragment type
Oxytocin 1	504.6	933.0	11	b8+
Oxytocin 2		285.2	15	y3+
IS Oxytocin 1	507.5	937.2	11	b8+
IS Oxytocin 2		289.1	21	y3+
Source parameters
Capillary voltage		3 kV
Cone voltage		30 V
T° source		150 °C
T° desolvation		600 °C

2.5. Matrix comparison

Two potential matrices were tested to determine the most suitable: (a) PBS with 0.1 % BSA, and (b) cow plasma diluted in PBS with 0.1 % BSA (2:8, v:v). The 10 calibrators were prepared as mentioned in 2.2, 2.4. Thirty-six human plasma samples were spiked with different concentrations of oxytocin and analyzed by LC-MS/MS. The calculated concentrations, following LC-MS/MS analysis and peak size comparison to the calibration curves data, ranged from 0.55 to 170 ng/L. Passing-Bablok regressions were then performed to compare the oxytocin human plasma samples concentrations calculated with both calibration curves. This regression has been chosen to assess if both calibration curves provide the same result for the same patient. A calibration curve was also prepared in human plasma to compare the different slopes obtained with the various matrices.

2.6. Method validation

Guidelines provided by the CLSI (CLSI-C62A) [27] were referenced for the design of the validation process. Validation samples (i.e., 1 ng/L, 2.5 ng/L, 5 ng/L, 7.5 ng/L, 12.5 ng/L, 25 ng/L, 50 ng/L and 75 ng/L), were prepared and subjected to analysis in four replicates over five consecutive days.

2.6.1. Selectivity

Given the substantial amino acid sequence homology (78 %) between vasopressin and oxytocin [28], it was necessary to confirm the chromatographic separation of these two proteins. A pool of human plasma samples was prepared, divided into 1350 µL aliquots in 2 mL protein LoBind Eppendorf tubes, processed in accordance with the sample preparation section, and injected into the mass spectrometer. The samples were analyzed using an LC-MS/MS method that included the transitions of oxytocin and the transitions of vasopressin, namely 542.7 > 328.3 and 542.7 > 757.2.

2.6.2. Linearity

The linearity of the method was evaluated following the Study Design A1 described in CLSI guideline EP06 [29]. As recommended, a pool of human EDTA plasma was spiked with oxytocin to obtain a 120 ng/L solution. Subsequently, this spiked plasma underwent dilution with varying volumes of blank sample to constitute samples of five concentrations: 120 ng/L, 96 ng/L, 72 ng/L, 48 ng/L and 24 ng/L. Samples were prepared in quintuplicate and analyzed in the same analytical run.

2.6.3. Calibration curve

The calibration curve was established by correlating the peak area ratio of each oxytocin standard to its SIL-IS concentration in ng/L. To assess linearity, linear, quadratic, and cubic regression models, each subjected to a weighted analysis (1/x²), were applied to the calibration curve. 1/x² weighting was used to positively discriminate low calibration points, as the normal level of oxytocin was expected to be at the low ng/L level. Linearity was considered acceptable when the difference between the nonlinear and linear fit was below or equal to 5 %.

2.6.4. Precision and accuracy

Each validation sample underwent extraction following the procedure detailed in the sample preparation section and was subjected to processing and analysis in four replicates over five consecutive days.

The single-nested Analysis of Variance (ANOVA) method was used to evaluate inter- and intra-run precision and accuracy, as stipulated by CLSI EP05 guidelines [30]. Coefficients of variation (CV, %) represented the levels of precision, while accuracy was represented by mean recovery (%).

2.6.5. Sample preparation recovery and matrix effects

To determine sample preparation recovery, spiking solutions of oxytocin and SIL-IS at a concentration of 2x10⁵ ng/L (high concentration) and 540 ng/L (low concentration) were prepared in H₂O, 5 % ACN and 0.1 % FA. In triplicate, 1300 µL of diluted cow plasma was spiked with 40 µL of the oxytocin solution. These samples were then subjected to extraction following the procedure outlined in the sample preparation section. During the reconstitution step, 40 µL of SIL-IS solution (2x10⁵ ng/L or 540 ng/L) and 40 µL of reconstitution solution were added to the extracts. As a reference, diluted cow plasma without spiked oxytocin was also extracted. The reconstitution solvent was a mixture of 40 µL of oxytocin (2x10⁵ ng/L or 540 ng/L) and 40 µL SIL-IS (2x10⁵ ng/L or 540 ng/L). The absolute recoveries were calculated using the following formula:

$$\frac{\frac{{\mathit{AUC}}_{\mathit{Oxy},S}}{{\mathit{AUC}}_{\mathit{SIL}-IS,S}}}{\frac{{\mathit{AUC}}_{\mathit{Oxy},Ref}}{{\mathit{AUC}}_{\mathit{SIL}-IS,Ref}}}\ast 100$$

Where AUC corresponds to the area under the curve, S refers to the sample extracted, Ref refers to the reference, Oxy to oxytocin and SIL-IS to the internal standard of oxytocin.

Matrix effects were determined by adding the same quantity of SIL-IS (1x10³ ng/L final concentration, to ensure a sufficient signal and a good peak shape) to six extracted human plasma samples and a solvent extract (H₂O without oxytocin). The matrix effect was calculated using the following equation and expressed as a percentage (%):

$$\frac{{\mathit{AUC}}_{\mathit{SIL}-IS,Human}}{{\mathit{AUC}}_{\mathit{SIL}-IS,H2O}}\ast 100$$

2.6.6. Carry over

Three consecutive injections of a standard solution at a concentration of 100 ng/L, corresponding to the highest point of the calibration curve, were performed, followed by the injection of a solvent solution (H₂O, 5 % ACN, 0.1 % FA) five times to determine carry over.

2.6.7. Sample stability

Sample stability was evaluated as follows:

Pre-extraction stability: Three samples (7.5 ng/L, 20 ng/L and 50 ng/L) were prepared and stored at 4 °C for varying periods ranging from one to five days. Subsequently, these samples were subjected to extraction and analysis to evaluate their stability.

Post-extraction stability: Another set of three samples was prepared, directly extracted, and analyzed over a span of five consecutive days. The samples were stored in the autosampler at 10 °C during this period.

Standard solution stability: A SIL-IS solution was analyzed over five consecutive days to evaluate its stability. The sample was stored in the autosampler at 10 °C during this period.

Storage at −80 °C: The same three samples were prepared and directly stored at −80 °C for one to five days before extraction and analysis.

2.6.8. LLOQ Determination

The LLOQ denotes the lowest concentration that can be consistently and accurately detected while meeting the stipulated requirements for precision and accuracy, namely 20 % as defined by CLSI guidelines. This value was determined by performing a serial dilution of a standard sample with PBS with 0.1 % BSA to a concentration of 0.5 ng/L. Each diluted sample was then processed and analyzed according to our method.

In order to assess the precision and accuracy at LLOQ level, 40 replicates underwent extraction and analysis over a 10-day period.

For reliable performance, the precision, expressed as the CV should not exceed 20 %, ensuring robustness and repeatability. Additionally, the trueness bias needs to be less than 20 %, ensuring accuracy within acceptable limits.

The limit of detection (LOD) is defined as the lowest amount of a substance that can be reliably distinguished from a blank sample. The LOD is determined by establishing the minimum concentration at which the signal to noise ratio (S/N) is greater than 3.

2.6.9. Measurement uncertainty

Throughout the validation process, four types of uncertainty were computed using e.noval 4.1 software (Arlenda, Liège, Belgium): (1) uncertainty of the bias, which is related to the bias of the method; (2) uncertainty that accounts for both the bias and the intermediate precision standard deviation; (3) expanded uncertainty corresponding to the uncertainty multiplied by a coverage factor ‘K’ (where ‘K’ equals 2), delimiting an interval within which the true value should appear with a 95 % confidence level; and (4) the relative uncertainty determined by dividing the expanded uncertainty by the anticipated or expected concentration, delimiting an interval (centered on the introduced value) within which the true value should appear with a 95 % confidence level [31].

Considering the absence of available data on the biological variation of oxytocin, a proactive decision was made to establish the Measurement Uncertainty (MU) goal as being below or equal to 15 % as stated in CLSI guidelines.

In addition to uncertainties, the β-expectation tolerance intervals were also determined. These intervals contain a proportion β of the individual values, such as results, of the population under investigation, thus describing the entire population. If β = 0.95, this means that, on average, 95 % of the future individual values of the population are included in the interval [32].

The β −expectation tolerance intervals and the accuracy profile were calculated using e.noval 4.1 software.

3. Results

3.1. Matrix comparison

Results of the passing-Bablok regression are shown in Fig. 4(a). The comparison between the calibration curves prepared in cow plasma diluted in PBS with 0.1 % BSA and PBS with 0.1 % BSA showed significant bias (p < 0.05). The regression equation obtained was: [PBS with 0.1 % BSA calibration curve] = 0.0446 (−0.0569 to 0.1202) + 0.9275 (0.9126 to 0.9651) [Cow plasma diluted in PBS-0.1 % BSA calibration curve]. Based on this analysis, PBS with 0.1 % BSA was selected for further use. Moreover, as seen in Fig. 4(b) a similar slope was obtained in the calibration curve realized in human plasma.

Fig. 4.

(a) Passing-Bablok regressions comparing the calculated concentration of 36 patients obtained with calibration curve prepared in cow plasma diluted in PBS with BSA 0.1% (2:8) and a calibration curve prepared in PBS with BSA 0.1%. (b) Comparison between calibration curves prepared in human plasma (green), cow plasma diluted in PBS with BSA 0.1% (2:8) (orange) and PBS with BSA 0.1% (blue).

3.2. Selectivity

Our method effectively achieved chromatographic separation of oxytocin and vasopressin, as demonstrated in Fig. 5. Oxytocin was detected with a retention time of 4.8 min, while vasopressin was characterized by a retention time of 3.8 min.

Fig. 5.

Chromatographic separation of oxytocin (orange) and vasopressin (blue) in human plasma.

3.3. Linearity

Table 3 shows that the calculated concentrations for the diluted samples fell within the expected accuracy range (i.e., < 15 %) confirming the linearity of the dilution process.

Table 3.

Evaluation of the method linearity.

Expected concentration (ng/L)	Calculated concentration (ng/L)	Accuracy (%)
120	115	96
96	89	92
72	76	106
48	55	114
24	25	105

3.4. Calibration curve

The linearity of the curve was confirmed as acceptable, with the difference between the nonlinear and linear fit results remaining below the threshold of 5 %, as shown in Table 4.

Table 4.

Linearity evaluation of the oxytocin calibration curve.

Expected concentration (ng/L)	Quadratic	Linear	Cubic	CV (%)
1	0.9	0.9	0.9	1
2	2,0	2.1	2.1	2
4	4.4	4.3	4.4	1
6	6.1	6.0	6.1	1
10	9.7	9.5	9.6	1
20	21.6	21.3	21.6	1
40	36.9	36.2	36.5	1
60	59.2	59.9	59.8	1
80	81.8	85.9	83.9	2
100	100.2	102.0	98.0	2

3.5. Precision and accuracy

Accuracy ranged from 94 % to 108 %, with intra-day CVs spanning from 1 % to 10 % and inter-day CVs ranging from 3 to 7 %. Table 5, Table 6 provide detailed inter- and intra-day CVs and accuracies obtained during the validation.

Table 5.

Inter and intra-day CV and absolute accuracy for oxytocin in PBS with BSA 0.1%.

Validation sample	Concentration (ng/L)	Day	Accuracy (%)	Intraday CV (%)	Interday CV (%)
1	1	1	103	5	6
		2	100	6
		3	102	5
		4	101	7
		5	97	10
2	2.5	1	103	5	6
		2	105	5
		3	97	7
		4	99	6
		5	97	4
3	5	1	97	1	4
		2	101	3
		3	105	2
		4	97	4
		5	100	5
4	7.5	1	100	1	4
		2	101	3
		3	100	3
		4	96	6
		5	101	4
5	12.5	1	102	1	3
		2	98	3
		3	101	2
		4	103	3
		5	102	1
6	25	1	103	3	6
		2	94	2
		3	98	6
		4	96	2
		5	105	1
7	50	1	105	2	4
		2	98	2
		3	102	3
		4	96	2
		5	105	1
8	75	1	106	1	4
		2	101	1
		3	101	5
		4	103	4
		5	99	2

Table 6.

Inter and intra-day CV and absolute accuracy for oxytocin in human plasma.

Validation sample	Concentration (ng/L)	Day	Accuracy (%)	Intraday CV (%)	Interday CV (%)
1	5	1	105	5	7
		2	108	8
		3	108	5
		4	104	8
		5	103	10
2	50	1	101	3	3
		2	97	3
		3	102	2
		4	102	3
		5	100	1
3	100	1	106	1	4
		2	95	2
		3	102	2
		4	101	3
		5	98	1

The accuracy profile generated by the e.noval software is represented in Fig. 6, where calculated values fell within the 15 % acceptance limits for each concentration level and within 20 % for the LLOQ.

Fig. 6.

Accuracy profile for oxytocin. The relative bias is represented by red lines, the β-expectation tolerance limits by dashed blue lines and the acceptance limits by dashed black lines. The series represent the five days of validation. The dots represent the relative error of the results and are plotted with respect to their targeted concentration.

3.6. Sample preparation recovery and matrix effects

At high concentration (i.e., 2x10⁵ ng/L), sample preparation recoveries ranged between 90 % and 96 %, and between 50 % and 70 % at low concentration (i.e., 20 ng/L), while the calculated matrix effects ranged between 18 % and 27 %.

3.7. Carryover

After injecting a 100 ng/L solution of oxytocin, no residual traces of this peptide were detected in subsequent analysis, confirming the absence of carryover, as illustrated in Fig. 7.

Fig. 7.

Carry over evaluation. (a) A 100 ng/L oxytocin (after SPE treatment) injection followed by (b) 5 blanks. The different injections a represented here by their TIC.

3.8. Stability

Regarding the plasma samples (pre-extraction stability) stored for one to five days at 4 °C, it was observed that all three concentrations showed degradation after one day. For extracted samples (post-extraction stability), concentrations of 20 ng/L and 50 ng/L remained stable (with CVs ≤ 15 %) over five days, contrary to the low concentration (i.e., 7.5 ng/L). The SIL-IS solution remained stable for up to four days at 4 °C after being homogenized prior to injection. As shown in Table 7, a CV egal or less than 15 % was obtained for all samples over this four day period. Additionally, when stored directly at −80 °C for one to five days, all three tested concentrations remained stable (CV ≤ 15 %). The results are detailed in Table 7.

Table 7.

Stability evaluation (CV,%) for oxytocin standards and SIL-IS over five days. CV’s greater than cut-off (15%, unstable) are in red.

	Day 2	Day 3	Day 4	Day 5
Pre-extraction
7.5 ng/L	33 %
20 ng/L	26 %
50 ng/L	15 %	30 %
Post-extraction
7.5 ng/L	28 %
20 ng/L	15 %	15 %	15 %	17 %
50 ng/L	13 %	12 %	13 %	15 %
SIL-IS
1000 ng/L	10 %	8 %	9 %	16 %
Stored −80 °C
7.5 ng/L	0 %	1 %	1 %	2 %
20 ng/L	1 %	2 %	6 %	5 %
50 ng/L	3 %	4 %	4 %	4 %

3.9. LLOQ

The serial dilution of the standard sample led to the decision to define 1 ng/L as LLOQ. The validation of the 1 ng/L sample demonstrated an average intra-day CV of 6.6 %, an inter-day CV of 6.3 %, and an average accuracy of 101 %, as shown in Table 5. This level was designated as the LLOQ. The response corresponding to the LLOQ is illustrated in Fig. 8.

Fig. 8.

Chromatographic representation of the LLOQ 1 ng/L (after SPE treatment in PBS with 0.1 % BSA matrix) for the transition m/z 504.5 −> m/z 933.04.

3.10. Measurement uncertainty

Relative uncertainties, ranging from 6 % to 14 %, were below 15 %. Detailed values are provided in Table 8. As shown in Fig. 6, the relative upper and lower β-expectation tolerance limits (%) were within the acceptance limits (+/- 15 % and +/- 20 % for LLOQ) for each concentration level.

Table 8.

Oxytocin measurement uncertainty.

Expected concentration (ng/L)	Uncertainty of the bias (ng/L)	Uncertainty (ng/L)	Expanded uncertainty (ng/L)	Relative expanded uncertainty (%)
PBS with 0.1 % BSA
1	0.02	0.08	0.14	14
2.5	0.04	0.15	0.31	12
5	0.07	0.23	0.46	9
7.5	0.07	0.31	0.61	8
12.5	0.11	0.36	0.71	6
25	0.52	1.44	2.88	12
50	0.90	2.39	4.79	10
75	0.94	3.01	6.03	8
Human plasma
5	0.08	0.36	0.72	14
50	0.52	1.66	3.33	6
100	1.83	4.82	9.65	9

4. Discussion

Here, we successfully developed a highly sensitive LC-MS/MS method for the accurate quantification of oxytocin in plasma. Unlike existing methods (Table 1), our approach achieved a remarkable lower limit of quantification of 1 ng/L without the need for 2D LC-MS/MS, nano LC-MS/MS or a HRMS. As previously noted by Franke et al. [22], the routine used of 2D-LC-MS/MS is not yet mainstream in clinical laboratories. Additionally, HRMS is expensive and is not readily available in all clinical laboratories.

In this work, we adapted a solid-phase extraction, initially developed by Zhang et al. [24]. Our sample preparation process took less than four hours for up to 44 samples. However, the number of samples could be increased without significant impact due to the use of a 96-well plate. This contrasts the method by Franke et al. [22], which includes an additional reduction/alkylation step for disulfide bridges before the solid-phase extraction (SPE). Furthermore, Franke et al. used Sep-Pak C18 cartridges for plasma treatment, which are not suitable for routine analysis of large batches of samples. Another method, salt-assisted liquid–liquid extraction (SALLE), was employed by Liu et al. [23]. However, their approach does not mention any guidelines for the validation process and does not enable reproducible quantification at low levels of oxytocin (i.e., 10 ng/L). Additionally, none of the previously developed methods have estimated the measurement uncertainty, a crucial factor for assessing the reliability and accuracy of measurement results.

Throughout our work, we used a single LC column during the entire validation process without any signal degradation, yielding consistently reproducible results. Moreover, a “mini” validation was also performed using human plasma. The validation outcomes confirm that our method is within the requirements of CLSI guidelines and attest to the robustness of our method confirming its suitability for clinical practice.

In terms of stability, we observed a significant reduction in oxytocin levels after spiking and prior sample preparation. As previously noted by various groups [22], [24], this decrease is related to the strong protein binding of oxytocin via its disulfide bridges. Therefore, it is crucial to store samples at −20 °C or −80 °C immediately after collection.

Unlike immunoassays, which are susceptible to cross-reactivity, our newly developed method showed no interference from vasopressin, despite their similar sequence.

However, it is important to acknowledge a significant drawback of our method: significant matrix effects with an average signal recovery of only 20 %. These issues likely arose from the fact that both the SPE plate sorbent and the stationary phase of the chromatographic column rely on dispersion interactions. As a result, there is no orthogonal separation of the compounds, increasing the risk of heavy coelution of unseparated compounds. Although we tested alternative SPE plates, including Oasis Max and Strata-X, none achieved full retention of oxytocin, with recovery rates below 30 %. Despite the high matrix factor, our method successfully reached an exceptionally low LLOQ. Another drawback of the method is the required plasma volume of 1350 µL, which may potentially exclude young children from further studies.

5. Conclusion

This work presents a robust and advanced LC-MS/MS method, offering high sensitivity and selectivity for the accurate quantification of oxytocin in plasma. Given the high prevalence of psychiatric disorders worldwide, this method could play a crucial role in deepening our understanding of oxytocin levels and their role in the etiology of these conditions. Additionally, it could be valuable in assessing the percentage of oxytocin released into the bloodstream following intranasal administration, thereby helping to optimize treatment dosages.

As oxytocin agonists and antagonists are currently under development, further investigation into their therapeutic potential is needed [14]. Endogenous oxytocin and stimulation of the oxytocin receptor are associated with growth, resilience, and healing [8]. Integrating studies on the effect of exogenous oxytocin with research on endogenous oxytocin levels could provide a more comprehensive understanding of oxytocin dynamics and its allostatic role across various physiological and psychological processes throughout life [33]. Future research should incorporate advances in molecular biology, polygenic risk factors, detailed clinical characterization, and interactions with the exposome and epigenetics to guide precision medicine. Enhancing research on the determinants of brain health and the individualized prevention of brain disorders is a fundamental priority.

CRediT authorship contribution statement

E. Grifnée: Writing – original draft, Validation, Methodology, Investigation, Conceptualization. A. Mackowiak: Validation. J. Demeuse: Writing – review & editing, Conceptualization. M. Schoumacher: Writing – review & editing. L. Huyghebaert: Writing – review & editing. W. Determe: Writing – review & editing. T. Dubrowski: Writing – review & editing. P. Massonnet: Writing – review & editing. S. Peeters: Writing – review & editing. G. Scantamburlo: Resources. E. Cavalier: Writing – review & editing, Supervision, Resources, Methodology, Conceptualization. C.Le Goff: Writing – review & editing, Supervision, Resources.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.