Graphene Oxide-Supported QuEChERS Extraction Coupled with LC-MS/MS for Trace-Level Analysis of Wastewater Pharmaceuticals

Featured Application

This study presents a novel sample preparation strategy combining modified QuEChERS with graphene oxide clean-up. This strategy can be applied to routine LC-MS/MS-based monitoring of pharmaceutical residues in wastewater to mitigate the impact of pharmaceuticals on aquatic ecosystems.

Abstract

Detecting pharmaceuticals in environmental matrices, particularly in wastewater, is crucial due to their potential environmental occurrence and unpredictable ecological and health-related consequences. These substances, often present in trace amounts, require highly sensitive and selective analytical methods for effective monitoring. A modified version of the QuEChERS (Quick, Easy, Cheap, Effective, Rugged, and Safe) method was evaluated to evaluate 18 pharmaceuticals and 2 metabolites in wastewater samples using liquid chromatography with tandem mass spectrometry (LC-MS/MS). The method’s performance was assessed using linearity, recovery, precision, limits of quantification (LOQ) and detection (LOD), and the matrix effect (ME). The final method was based on acetonitrile, Na₂EDTA, citrate buffer, and graphene oxide (GO). Finally, the calibration curves prepared in acetonitrile and the matrix extract showed a correlation coefficient of 0.99. Most of the compounds had LOQ values lower than 0.5 μg⋅mL⁻¹. Recoveries were achieved in the 70–98% range, with RSD lower than 13%. GO allowed the elimination of the ME, which occurred in the range of −11% to 15%. The results indicate that a low-cost and straightforward method is suitable for routinely monitoring pharmaceuticals in wastewater, which is crucial for minimizing the impact of pollutants on aquatic ecosystems.

1. Introduction

The presence of pharmaceuticals in the environment is an alarming problem, yet their widespread use is essential for treating diseases in humans and animals [1]. Changes in clinical practice, a continuing increase in population, and higher incidences of chronic and age-related diseases are leading to a raised need and consumption of pharmaceuticals. This trend is evident in Organisation for Economic Co-operation and Development (OECD) countries—between 2000 and 2019, the consumption of antihypertensive drugs increased by an average of 65%, antidiabetic and antidepressant drugs doubled, while the increase in the uptake of lipid-modifying medicines was as much as 400% [2]. Furthermore, the expanding proportion of people aged 65 and older globally leads to aging populations in most countries, a key driver for the pharmaceutical market. The market has experienced significant growth since the beginning of the century, with linear growth and revenues of USD1.61 trillion worldwide in 2023. It is expected to grow further, with a compound annual growth rate (CAGR) of 7.6% from 2023 to 2030 [3].

Increased pharmaceutical production entails the risk of their release into the environment, both at the production and distribution stage and due to inappropriate use, disposal of unused or expired medicines, and inefficient treatment of wastewater and landfill leachate. As a result, active substances contained in medicines can accumulate in the environment, posing a potential threat to ecosystems and public health [4].

Furthermore, pharmaceuticals are processed in the body once they are taken in, and then between 30% and 90% of the dose taken is excreted unchanged or metabolized, mainly in feces or urine [5,6]. They end up in treatment plants and domestic wastewater treatment systems in these forms and subsequently enter ecosystems. Pharmaceuticals resist degradation, persist for a long time in the environment, and become transported along food chains. Recent work constantly indicates the presence of pharmaceuticals in soil [3,7,8,9,10,11], surface water [10,12,13], groundwater [13,14], and even drinking water [15,16,17,18,19,20]. Pharmaceuticals are also taken up by soil organisms [21] and plants [22,23]. Their ubiquity raises concerns about possible environmental and human health risks [24], highlighting the need to monitor wastewater, the overriding source of pharmaceuticals in the environment.

Wastewater treatment is a crucial component in protecting the environment and humans from the presence of pharmaceuticals in ecosystems and the following consequences. Nevertheless, conventional treatment methods do not entirely remove some drug residues, potentially leading to their presence in wastewater leaving treatment plants [1]. Hence, there is an urgent need for reliable and detailed methodologies to determine concentrations of pharmaceuticals in wastewater before they enter the environment and to realistically determine the effectiveness of the technology used at the treatment plant.

Consequently, developing efficient analytical methods, including precise sample preparation, is becoming essential for effectively monitoring pharmaceutical residues in wastewater and evaluating treatment efficiency. Sample preparation is a critical step, affecting the quality of the results obtained by removing interferents from the matrix, stabilizing analytes, and ensuring compatibility with downstream analytical techniques. Many procedures have been developed for the preparation of aqueous samples, including wastewater, such as solid-phase extraction (SPE) [25,26,27], as well as microextraction techniques: dispersive liquid–liquid microextraction (DLLME) [28,29,30,31], liquid microextraction using hollow fiber (HF–LPME) [32,33,34], ultrasound-assisted emulsification microextraction (USAEME) [35,36,37], and single-drop microextraction (SDME) [38,39,40] have been reported. An alternative, less researched strategy for determining pharmaceuticals in environmental samples is using QuEChERS-based extraction methods (fast, easy, cheap, effective, robust, and safe) [41].

The QuEChERS method was initially designed to extract pesticide residues from fruits and vegetables [42]. The method was subsequently applied to the analysis of other food matrices, such as cereals [43,44]; legumes [44]; oils [45]; and environmental samples, e.g., soil [1,46,47,48]. Due to its simplicity and “low bias” toward specific groups of compounds, the technique was also developed to study residues of alkylphenols, bisphenol A [49], mycotoxins [50,51], pyrrolizidine alkaloids [52], polycyclic aromatic hydrocarbon (PAHs) [53], and surfactants [54], including per- and polyfluoroalkyl substances (PFASs) [55] in a variety of matrices. Due to the possibility of overcoming the complexity of the wastewater matrix and the great diversity of pharmaceuticals in physicochemical properties, modifications of the QuEChERS method appear to be a promising technique for extracting these compounds from wastewater. The technique avoids several drawbacks of traditional SPE, such as high solvent consumption, lengthy procedures, and poor selectivity in complex wastewater matrices. When combined with graphene oxide (GO) as a clean-up sorbent, the method enhances overall extraction efficiency and limits matrix effects, making it a more effective tool for detecting pharmaceutical residues in wastewater.

This work aims to improve the QuEChERS technique for simultaneously determining a range of pharmaceuticals in wastewater by improving sample extraction and clean-up procedures. Improving the accuracy and efficiency of the analysis for the detection of drugs in wastewater will support the monitoring of their levels in wastewater treatment plants and, therefore, contribute to the development of effective methods to control the efficiency of these facilities. Effective surveillance is crucial for adequate water quality management, helping to improve wastewater treatment methods and minimizing pollutants’ impact on aquatic ecosystems. In addition, it will provide a low-cost and straightforward tool to track the type and quantity of substances collectively consumed by the populations served by the treatment plants.

2. Materials and Methods

2.1. Materials and Reagents

Acetonitrile, acetone, methanol, ethyl acetate, formic acid (all LC-MS grade), ammonium formate (NH₄HCO₂; ≥99%), and ethylenediaminetetraacetic acid disodium salt dehydrate (Na₂EDTA; ≥99%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Ultrapure water (LC grade 18 Ω cm) was obtained from a MilliQ water purification system (Millipore Ltd., Bedford, MA, USA).

Pre-weighed QuEChERS kits containing analytical-grade reagents—magnesium sulfate, sodium chloride (NaCl), trisodium citrate dihydrate (Na₃C₆H₅O₇⋅2H₂O), and disodium hydrogen citrate sesquehydrate (Na₂HC₆H₅O₇⋅1.5H₂O) salt—were procured from Agilent Technologies (Santa Clara, CA, USA); pre-weighed kits containing magnesium sulfate (MgSO₄) and sodium acetate (CH₃COONa) were obtained from Supelco (Bellefonte, PA, USA).

For clean-up methods, pre-weighted sorbent mixtures PSA/C18/MgSO₄ (primary secondary amine/octadecylsilane/anhydrous magnesium sulfate) and PSA/GCB/MgSO₄ (primary secondary amine/graphitized carbon black/anhydrous magnesium sulfate) kits were purchased from Perlan Technologies (Santa Clara, CA, USA). Chitosan from shrimp shells, Fe₃O₄ magnetic nanoparticles (Fe₃O₄MNPs), and graphene oxide dispersion in H₂O (GO) were obtained from Sigma-Aldrich (Steinheim, Germany). The GO dispersion, at a concentration of 2 mg/mL, is chloride-free (purified by dialysis), consists of monolayer sheets, and has an average sheet diameter of less than 10 µm.

For the preparation of the matrix, i.e., the model wastewater according to PN-72/C-04550.09, the following components were used (all with purity ≥ 98%): casein peptone, enriched dry broth, ammonium chloride (NH₄Cl), calcium chloride (CaCl₂), monobasic potassium (KH₂PO₄), dibasic potassium phosphate (K₂HPO₄), and sodium acetate (CH₃COONa) from Sigma-Aldrich (Steinheim, Germany).

2.2. Standards

Analytical standards (purity ≥ 95%): atenolol, bisoprolol, candesartan cilexetil, ciprofloxacin, duloxetine hydrochloride, doxycycline hydrochloride, escitalopram oxalate, fluoxetine hydrochloride, indapamide, levonorgestrel, metformin hydrochloride, metoprolol, propafenone hydrochloride, ramipril, sertraline hydrochloride, torasemide anhydrous, valsartan, and venlafaxine hydrochloride and its metabolites—venlafaxine O-desmethyl and venlafaxine N-oxide—were purchased from Sigma-Aldrich (Steinheim, Germany). The following substances were used as internal standards (IS): atenolol-d7 (≥97%) and metformin-d6 (≥95%), purchased from Sigma Aldrich (Steinheim, Germany). An overview of the physicochemical properties of the studied target drug active substances is given in

Table 1. Physicochemical profiles of target drug active substances.

Compound and CAS Number	Molecular Formula	Structural Formula	Molar Mass [g⋅mol−1]	Bioavailability	Lowest PNECs Freshwater [µg⋅L−1]	pKa	Water Solubility [mg⋅mL−1]	logP	Melting Point [°C]
Angiotensin II receptor antagonists
candesartan 139481-59-7	C24H20N6O3		440.5	15%	100	3.44/1.5	>0.01	6.1	183–185
valsartan 137862-53-4	C24H29N5O3		435.5	25%	560	4.7	0.02	1.5	116–117
Angiotensin-converting enzyme inhibitors
ramipril 87333-19-5	C23H32N2O5		416.5	28%	1000	3.75/5.2	0.04	2.9	109
Antiarrhythmics
propafenone 54063-53-5	C21H27NO3		341.5	nd	0.85	14.09/9.63	>0.01	3.2	173
Antibiotics
ciprofloxacin 85721-33-1	C17H18FN3O3		331.3	70%	0.064	6.1	1.35	0.3	255–257
doxycycline 564-25-0	C22H24N2O8		444.4	~100%	2	3.09	50	0.6	201
Antidepressants
duloxetine 116539-59-4	C18H19NOS		297.4	32–80%	0.43	9.7	>0.01	4	164
escitalopram 128196-01-0	C20H21FN2O		324.4	~80%	2.68	9.5	>0.01	1.3	147–152
fluoxetine 54910-89-3	C17H18F3NO		309.3	60–80%	0.1	9.8	>0.01	4.1	193–197
sertraline 79617-96-2	C17H17Cl2N		306.2	44%	0.0094	9.16	3.8	5.1	245–246
venlafaxine 93413-69-5	C17H27NO2		277.4	~42%	0.88	14.42/8.91	572	2.7	240
Antidiabetics
metformin 657-24-9	C4H11N5		129.2	50–60%	160	12.4	1.38	−2.6	223–226
Beta-blockers
atenolol 29122-68-7	C14H22N2O3		266.3	50–60%	150	9.6	0.7	0.2	158–160
bisoprolol 66722-44-9	C18H31NO4		325.5	>90%	92	9.5	0.07	2.2	100–103
metoprolol 51384-51-1	C15H25NO3		267.4	50–70%	8.6	9.7	0.40	2.2	120
Diuretics
indapamide 26807-65-8	C16H16ClN3O3S		365.8	nd	1.29	8.8	0.03	2.5	160–162
torasemide 56211-40-6	C16H20N4O3S		348.4	80–90%	nd	7.1	0.06	3.4	163–164
Hormones
levonorgestrel 797-63-7	C21H28O2		312.5	85–100%	0.00001	17.91/−1.5	>0.01	3.8	232–239

nd—no data.

.

The stock standard solutions of pharmaceuticals (at around 1000 μg⋅mL⁻¹) were prepared by dissolving an accurate weighted amount of reference standards in an appropriate solvent (acetone, acetonitrile, methanol, or an appropriate mixture). The working standard solutions of multiple compounds, in concentrations of 0.001, 0.005, 0.010, 0.050, 0.100, 0.500, and 1.000 μg⋅mL⁻¹, were prepared via serial dilution of the stock solution of individual pharmaceuticals in acetonitrile. They were used for spiking and preparing solvent and matrix-matched calibration curves for the pharmaceuticals in the solvent and matrix. A solution of the internal standards (atenolol-d7 and metformin-d6) was prepared in acetonitrile (0.5 μg mL⁻¹) and used to prepare working calibration solutions and during the sample spiking procedure. All stock and working standard solutions were protected from direct light and stored in dark glass bottles at 4 °C until analysis.

2.3. Model Wastewater

The procedure for optimizing pharmaceutical extraction and extract clean-up was carried out on model (synthetic) wastewater. They were prepared to reflect the qualitative composition of actual municipal wastewater. To this end, enriched broth (152 mg⋅L⁻¹) and casein peptone (226 mg⋅L⁻¹) were used to reflect the source of carbon compounds and organic nitrogen; NH₄Cl (20 mg⋅L⁻¹) was used as a source of ammonium nitrogen, and KH₂PO₄ (16 mg⋅L⁻¹) and K₂HPO₄ (40 mg⋅L⁻¹) were used as a source of phosphorus. CH₃COONa (212 mg⋅L⁻¹) was added to increase COD. The model wastewater was prepared using tap water filtered on an activated carbon column to preserve macro- and micronutrients and remove chlorine and organic pollutants [56]. Their BOD and COD values were approximately 300 mg⋅L⁻¹ and 500 mg⋅L⁻¹, respectively. The model wastewater was stored at −20 °C until analysis.

2.4. Sample Preparation Extraction and Clean-Up

2.4.1. QuEChERS-Based Extraction

The extraction of pharmaceuticals from wastewater was based on modified QuEChERS protocols. It was carried out in ten variants, each variant in three repetitions. The experimental details of the QuEChERS protocols tested are summarized in

Table 2. Details of the procedures tested for the QuEChERS-based extraction of pharmaceuticals from model wastewater.

Trial	Extraction Solvent	Buffer	Additives
1	acetonitrile	-	-
2		acetate	-
3		citrate	-
4		-	formic acid 1
5		-	0.2 M Na2EDTA
6		citrate	0.2 M Na2EDTA
7	ethyl acetate	-	-
8		citrate	-
9		-	0.2 M Na2EDTA
10		citrate	0.2 M Na2EDTA

¹ Previously added to the acetonitrile to achieve a concentration of 1% in solvent.

.

The final procedure is outlined below. In a 50 mL PTFE centrifuge tube, 10 mL of wastewater was placed. Subsequently, 10 mL 0.2 M Na₂EDTA in H₂O and 50 µL internal standard (0.5⋅μg mL⁻¹) were added and shaken vigorously. Then, 10 mL of acetonitrile was added and mixed for 1 min by hand, followed by 1 min on a Vortex shaker. The sample was placed in a freezer (−20 °C) for 15 min, and after that time, the pre-weighed citrate buffer kit was added. The sample was again shaken for 1 min by hand and 1 min on a Vortex shaker. The tube was then centrifuged for 5 min at 4500 rpm. A total of 5 mL of supernatant was transferred to a 15 mL PTFE tube and frozen for 30 min (−60 °C). After defrosting, the extract was filtered through a 0.45 µm PTFE filter, transferred into the glass autosampler vial, and analyzed via LC-MS/MS.

2.4.2. Clean-Up

The clean-up process was optimized for the most favorable extraction variant regarding recovery and matrix effect (i.e., the variant described in the previous point).

To the defrosted extract, as described in Section 2.4.1, one of five sorbents was added: pre-weighted PSA/C18/MgSO₄, PSA/GCB/MgSO₄, 200 mg of chitosan, 1 mL of GO, or 1 mL of Fe₃O₄MNPs, then vortexed for 1 min. Then, the extract was centrifuged for 10 min at 4500 rpm, filtered through a 0.45 mm PTFE filter, transferred into the glass autosampler vial, and analyzed via LC-MS/MS.

2.5. LC-MS/MS Conditions

An Eksigent Ultra LC-100 (Eksigent Technologies, Dublin, CA, USA) liquid chromatography system was operated at a flow rate of 0.4 mL min⁻¹ without split using a Kinetex C18 100 mm × 2.1 mm, 2.6 µm (Phenomenex, Torrance, CA, USA) analytical column, maintained at 40 °C during the experiments. The volume injected into the LC-MS/MS system was 10 μL. The binary mobile phase consisted of water with 0.2% formic acid and 5 mM ammonium formate (phase A) and methanol with 0.2% formic acid and 5 mM ammonium formate (phase B). The gradient elution starts at 99% A and 9% B, rising linearly to 40% A and 60% B in 1.5 min. In the next step, it rose linearly to 5% A and 95% B in 4.5 min and was held for 2.0 min. After ramping, the mobile phase composition was returned to the initial condition in 1 min, and this was held for 2.5 min for re-equilibration.

System MS/MS 6500 QTRAP (AB Sciex Instruments, Foster City, CA, USA) was used for mass spectrometric analysis, equipped with an electrospray ionization source (ESI). The capillary voltage was maintained at 5000 V for the positive ion mode, and the temperature of the turbo heaters was set at 400 °C. The nebulizer gas (GS1), auxiliary gas (GS2), and curtain gas (CUR) (nitrogen) were used at a pressure of 50, 40, and 35 psi, respectively. Nitrogen was also used as the nebulizer and collision gas. All pharmaceuticals were detected in the multiple reaction monitoring mode (MRM). For each compound, the precursor ion and two product ions were determined. One product ion was used for quantification, and one was used for qualification. The MRM transitions for the pharmaceuticals are given in

Table 3. MRM parameters for selected pharmaceuticals analyzed using LC-MS/MS.

Pharmaceutical or Metabolite	Retention Time (min)	Quantification				Confirmation
		MRM Transition m/z	EP (V)	CE (V)	CXP (V)	MRM Transition m/z	EP (V)	CE (V)	CXP (V)
Atenolol	2.35	267.3 > 267	5	35	2	267.3 > 133	2	39	4
Bisoprolol	3.7	326 > 116	10	27	8	326 > 74	10	43	12
Candesartan	5.4	441.2 > 63.2	10	74	17	441.2 > 207.2	12	54	10
Ciprofloxacin	3.05	332 > 288	7	25	8	332 > 314	7	29	8
Doxycycline	5.25	445.1 > 428	10	29	13	445.1 > 321	10	42	13
Duloxetine	4.05	298.1 > 44.1	3	33	4	298.1 > 154.2	3	11	4
Escitalopram	2.2	325.1 > 143	15	53	10	325.1 > 109.1	15	30	10
Fluoxetine	4.25	310.1 > 148.1	11	15	4	310.3 > 91.2	5	25	2
Indapamide	3.55	366.1 > 132.2	10	20	7	366.1 > 147.1	12	18	5
Levovorgestrel	4.75	313.3 > 245.3	10	25	12	313.3 > 96.3	8	28	10
Metformin	1.55	130.1 > 71	4	32	2	130.1 > 60	4	19	2
Metoprolol	2.75	268 > 191	10	27	12	268 > 145	10	35	4
Propafenone	3.95	342.4 > 116.1	8	32	7	342.4 > 98.1	12	29	6
Ramipril	4.2	417.2 > 234.2	5	30	3	417.2 > 117.1	5	57	2
Sertraline	4.75	306.1 > 159.1	7	35	4	306.1 > 275.1	5	17	4
Torasemide	3.6	349.1 > 264.1	5	25	3	349.1 > 168.1	5	66	2
Valsartan	4.65	436 > 235.2	7	37	11	436 > 291.2	7	23	6
Venlafaxine	3.55	278.2 > 58.1	8	37	4	278.2 > 260.2	8	17	4
Venlafaxine N-oxide	3.3	294 > 121	4	30	12	295 > 178	6	24	16
Venlafaxine O-desmethyl	3.1	264.1 > 58.1	11	31	4	264.1 > 246.4	11	17	4

Q1 and Q2 mass-to-charge ratios (m/z), retention times (RT), entrance potential (EP), collision energy (CE), and collision cell exit potential (CXP).

.

2.6. Method Validation and Matrix Effect

The developed method was validated using model wastewater to assess its linearity, recovery, precision, limit of quantification (LOQ), limit of detection (LOD), matrix effects (ME), and uncertainty (U).

The method’s linearity was assessed by analyzing a series of standard samples with six different concentrations, ranging from LOQ to 1.0 μg⋅mL⁻¹, in both pure solvent and matrix extracts of model wastewater over three consecutive days. The limit of quantification (LOQ) was determined as the lowest spiking level, validated by satisfactory recovery values (70–120%) and RSD (≤20%). The limits of detection (LOD) were calculated based on a signal-to-noise ratio of 3.

During recovery testing, model wastewater samples free of pharmaceuticals were fortified with appropriate volumes of standard pharmaceutical solutions at concentrations of 0.001, 0.01, and 0.1 μg⋅mL⁻¹. After a one-hour equilibration period, the samples were subjected to the previously described procedure. For each concentration level, five replicates were analyzed. Precision was determined as the relative standard deviation (RSD%) for each spiking level. Recoveries were deemed acceptable when falling within the range of 70–120%, with an RSD of no more than 20%.

The matrix effect (ME) was estimated by comparing the slopes of six points on the matrix-fitted calibration curves and the solvent-fitted calibration curves according to the following formula:

$$\%ME=\left({\displaystyle \frac{slop{e}_{matrix}}{slop{e}_{solvent}}}-1\right)\cdot 100$$

(1)

Depending on the percentage of ME, different matrix effects could be observed: negative—high signal suppression (≥50%) and moderate signal suppression (20 to 50%); no matrix effect (−20 < ME < 20%); positive—moderate signal enhancement (20–50%) and high signal enhancement (≥50%) [57,58].

The measurement uncertainty was estimated basing on the data obtained in the validation study. The relative expanded uncertainty was calculated by using the coverage factor k = 2 at the confidence level of 95% [59].

3. Results and Discussion

3.1. LC–MS/MS Optimisation

A review of relevant literature and the authors’ own experience indicated that liquid chromatography coupled with tandem quadrupole mass spectrometry (LC-MS/MS) in the multiple reaction monitoring (MRM) mode is well-suited for the determination of pharmaceutical residues with diverse physicochemical properties [60,61,62,63,64,65,66].

In the initial phase of the study, detection parameters were selected and optimized. For this purpose, standard solutions of individual pharmaceuticals at a concentration of 0.1 µg⋅mL⁻¹ were introduced into the mass spectrometer to establish appropriate operating conditions. The first aspect evaluated was the optimal ionization technique—electrospray ionization (ESI) or atmospheric pressure chemical ionization (APCI). Across all tested compounds, APCI produced a significantly lower signal intensity (in terms of peak height and area), which led to the selection of ESI due to its superior signal response.

The ionization mode (positive or negative) was also examined. The positive mode consistently yielded better signal intensity for all analytes, although negative ionization was found to be effective for acidic compounds such as candesartan, valsartan, and ramipril. The addition of ammonium formate to the standard solutions proved to be a critical factor, significantly enhancing signal intensity by facilitating ion formation.

The full-scan mode was initially used to identify the most intense precursor-to-product ion transitions for each compound. These transitions were then employed in the MRM mode, with two transitions selected per analyte to ensure reliable qualitative identification. In each case, protonated molecules ([M+H]+) were chosen as the precursor ions.

Subsequently, fragmentation of the selected ions was optimized by adjusting detector parameters specific to each analyte, including the declustering potential, collision energy, and ion exit potential from the collision cell. Fragmentation spectra were recorded at varying collision energies. Lower energy values favored the formation of higher-mass fragments, while increased energy led to more abundant low-mass ions and the disappearance of heavier ones. For each MRM transition, the collision energy that provided the highest sensitivity was selected. Specific values of tested parameters are presented in Table 3.

Determining the optimal parameters of ion sources, including curtain gas pressure, spray support pressure, auxiliary pressure, temperature, and voltage applied to the electrode, was an important part of the work of selecting MS/MS detection conditions. It was particularly challenging to select the aforementioned parameters in the case of the analysis of a mixture of 20 pharmaceuticals.

Compounds with diverse physicochemical features (diverse lipophilicity, acid–base nature, and different types and quantities of functional groups in molecule), deciding their retentive properties, were included in the composition of the separated mixtures.

The mixtures subjected to separation included compounds with varying physicochemical properties—such as differences in lipophilicity, acid–base character, and the presence of diverse functional groups—all influencing their chromatographic retention behavior.

Optimal chromatographic conditions were established based on key analyte characteristics, including molecular weight, logP, and pKa values. During the optimization process, several variables were evaluated: the type of column packing, its temperature, the mobile phase composition and pH, the flow rate, and the elution program. The primary objective was to achieve high separation efficiency, sensitivity, and reproducibility within a short analysis time (under 10 min). Although full separation of co-eluting analytes is not strictly required in MRM mode, improving the resolution reduces matrix interferences and enhances both sensitivity and selectivity. Additionally, the differences in retention times helped confirm analyte identity.

Choosing the appropriate stationary phase was a crucial step in developing the reversed-phase separation method. For this purpose, tests were performed on a series of columns with diverse parameters (length, diameter, grain size, and polarity) specifically designed for ultra-efficient liquid chromatography. The best results were obtained with a core–shell Kinetex C18 column (100 mm × 2.1 mm, 2.6 µm). A fixed, non-porous silicone core is the packing of this column, on which a porous silica layer is deposited, and a monolayer of dimethyl-n-octadecyl silane is immobilized on its surface. Such a phase is “endcapped” by means of the appropriate reagent to deactivate the silica as much as possible. This configuration ensured high efficiency and a sharp peak resolution.

The composition of the mobile phase also had a significant impact on chromatographic performance. Water served as the main component, while methanol and acetonitrile were tested as organic modifiers. Both solvents provided good retention, but methanol was chosen due to its lower cost. Adding acidic modifiers—acetic or formic acid—enhanced analyte retention, with formic acid proving more effective and thus being selected for final use. Lowering the mobile phase pH was found to improve peak shape, increase selectivity, and enhance sensitivity.

Gradient elution was necessary to effectively separate the broad range of pharmaceuticals. Adjustments to the gradient profile—such as modifying the initial proportion of the organic solvent—helped improve resolution. In early elution stages, weakly retained compounds (e.g., atenolol and metoprolol) migrated rapidly due to their low logP values. Shortening the gradient duration for the organic phase allowed better peak shapes for hydrophobic compounds and reduced the analysis time without compromising efficiency.

The developed chromatographic method was characterized by satisfactory reproducibility of retention times regardless of the number of determined pharmaceuticals or their physicochemical properties. Moreover, a short time of analysis, and thus low cost and solvent consumption, are fundamental advantages of the developed methodology.

3.2. Extraction and Clean-Up Optimisation

A significant step in this research was to develop a method for isolating and purifying pharmaceuticals from wastewater samples. Wastewater represents a complex matrix for sample preparation in assessing organic pollutants due to the synergy of chemical, physical, and biological factors. On the other hand, pharmaceuticals, as a group of pollutants, show great structural diversity. Different functional groups, such as esters, carboxylic acids, ketones, and amides, affect many of their characteristics, including stability, reactivity, and water solubility, significantly complicating the sample preparation process [67]. The main objective of this work was to develop a QuEChERS procedure that would shorten the complex and time-consuming extraction and clean-up processes and simultaneously provide good analyte recoveries and low ME. Model wastewater was enriched with known concentrations of pharmaceuticals to develop the method. The composition of the extraction medium was optimized by selecting organic solvents, buffer mixtures, and additives in formic acid and 0.2 M Na₂EDTA. The clean-up process was then optimized due to the complexity of the matrix and the consequent need for proper purification from various co-extracted interfering substances.

3.2.1. Extraction

Various modifications of the QuEChERS technique were used to isolate pharmaceuticals from wastewater. Depending on the variant, 10–106% recoveries were obtained for all compounds tested (Figure 1). Recovery values above 100% may result from matrix effects that cause signal enhancement, as well as inherent analytical variability. Such occurrences are common in complex matrices like wastewater and are accounted for in the method validation. The original procedure involves using acetonitrile as the extraction solvent (Figure 1). The original procedure involves the use of acetonitrile as the extraction solvent [42,68,69]. This variant (trial 1) yielded 25–88% recoveries, with satisfactory recoveries (above 70%) obtained for only seven compounds. The average recovery for this variant was 60%.

Acidic conditions can accelerate the desalting process and increase the stability of compounds in acetonitrile, as demonstrated for pesticides [70]. Furthermore, at acidic pH, the anionic sites of the matrix and analytes become protonated, leading to electrostatic repulsion between them, increasing the extraction efficiency [71]. Acidification of acetonitrile with formic acid (trial 4) significantly improved the recoveries of fluoxetine (from 62% to 72%), indapamide (from 67% to 85%), and ramipril (from 46% to 75%). However, it negatively affected the recovery of valsartan (from 74% to 59%). The use of ethyl acetate instead of acetonitrile (trial 7) improved the recoveries of ciprofloxacin (from 61% to 79%), fluoxetine (from 62% to 100%), indapamide (from 67% to 107%), torasemide (from 68% to 80%), and venlafaxine-O-desmethyl (from 58% to 75%).

The choice of buffers directly impacts the recovery rates of pharmaceuticals during extraction—it allows pH adjustment to ensure that as many compounds as possible are obtained in the neutral state. This procedure minimizes the presence of charged forms that could increase their affinity for water during drug extraction [72]. In the QuEChERS technique, there are two different standards regarding the type of buffer [68,73]: the US standard AOAC Official Method 2007.01 [74], which includes the use of an acetate buffer (pH about 4.8), and the European Committee for Standardisation (CEN) Standard Method EN 15662 [75], which includes the use of a citrate buffer (pH about 5–5.5), weaker in terms of ionic strength. For their comparison, an extraction with acetonitrile was performed. The acetate buffer-based method (trial 2) yielded satisfactory recoveries for only two beta-blockers (atenolol and metoprolol) and an antiarrhythmic drug (propafenone). The use of citrate (trial 3) yielded about twice the recovery of levonorgestrel (but still less than 70%) and fluoxetine, and nearly three times better for indapamide compared to the acetate buffer version (trial 2). No significant difference in recoveries (greater than 5%) was observed for bisoprolol, doxycycline, escitalopram, duloxetine, metoprolol, ramipril, sertraline, torasemide, and valsartan. However, the mean recovery rate for acetonitrile extraction (trial 3) with citrate buffer was higher compared to the version without buffer (trial 1) (67 and 60%, respectively).

The use of EDTA in the extraction of pharmaceuticals allows higher recoveries to be achieved by increasing the ionic strength, potentially increasing their solubility in the organic solvent [1]. Due to their strong chelating properties, EDTA and its salts can interact more strongly with metals in the matrix and thus exchange with analytes at active sites or mask adsorption sites [71]. Therefore, adding EDTA or its salts in the extraction process prevents the complexation of analytes with cations such as Mg²⁺ or Ca²⁺. Improvements in extraction efficiency by preventing chelate formation have been observed for fluoroquinolone and tetracycline compounds—for example, [76,77]. On the other hand, using EDTA may reduce the efficiency of the clean-up step [68].

In some cases, the presence of Na₂EDTA adversely affected pharmaceutical recoveries. Ethyl acetate extraction in the presence of Na₂EDTA (trial 9) yielded significantly lower recoveries of atenolol, ciprofloxacin, doxycycline, duloxetine, fluoxetine, and indapamide compared to the procedure without the chelating agent (trial 7) (mean recoveries of 64 and 70%, respectively). In the case of acetonitrile extraction, an adverse effect of the presence of Na₂EDTA (trial 5) was observed for only two of the three beta-blockers tested—atenolol and metoprolol—and sertraline, an SSRI antidepressant. A possible rationale for this is that when Na₂EDTA is in excess, it chelates metals and organic compounds [78].

Using the acetonitrile—citrate buffer—Na₂EDTA combination (trial 6) resulted in the best recoveries, 70–99% (average of 85%). The recoveries of bisoprolol, candesartan, and venlafaxine N-oxide in this variant were satisfactory (above 70%). Nevertheless, this variant had relatively high RSD values, as high as 22% for levonorgestrel, 24% for duloxetine, and 29% for metformin (mean of 14%).

The matrix effect, directly related to the use of the MS/MS technique, is a significant factor affecting the quality of the results. The presence of co-extracted substances in the sample can interfere with the ionization process at the spectrometer interface, leading to increased or decreased signal intensities of the analytes. Consequently, this can result in falsification of quantitative results [79]. The use of the described QuEChERS protocols was associated with the presence of substances in the extract, causing strong attenuation or amplification of the analyte signals of pharmaceuticals (ME between −89 and 79%). ME values for trial 6 were the most promising (ME between −28 and 37%); for 65% of the compounds, a satisfactory ME value (between −20 and 20%) was achieved (Figure 2). An additional clean-up step based on using different sorbents was introduced to eliminate or reduce ME.

3.2.2. Clean-Up

The clean-up step is necessary to reduce the ME, which strongly influences recoveries [80], but can also result in loss of analytes [68]. As indicated earlier, samples were previously frozen before the actual purification, which is the most common method for direct clean-up using the QuEChERS technique. This procedure mainly allows the removal of fats from the sample [81], which in the effluent can be up to 2000–5000 mg⋅L⁻¹, depending on the origin [82,83,84]. Compared to the melting point of fats (usually below 40 °C), the melting point of pharmaceuticals is generally high (Table 1)—it is estimated that for 78.9% of pharmaceuticals, it is in the range of 80–240 °C. In comparison, for more than 90%, it is higher than 50 °C [85,86]. Due to the differences in melting point, the extracted pharmaceuticals are still dissolved in the organic solvent. At the same time, the frozen co-extracted fat can be easily separated from the extracts via centrifugation [81].

To minimize ME during the analysis of pharmaceuticals in wastewater, the effectiveness of different sorption materials used to purify extracts—PSA/C18/MgSO₄, PSA/GCB/MgSO₄, chitosan, GO, and Fe₃O₄MNPs—was assessed. Without a clean-up step, 35% of pharmaceuticals showed signal suppression or enhancement, confirming the need for appropriate sample purification.

GO had a negligible effect on the recovery values (Figure 3); they were in the 70–98% range when used. It also reduced RSD values to 3–13%, and provided satisfactory ME values, ranging from −11% to 15%. The usefulness of GO has previously been demonstrated in determining 52 veterinary drugs in mutton, using a reduced graphene oxide-coated melamine sponge. The material enabled ME values ≤ ±13% to be obtained in the 10–500 μg·kg⁻¹ range [87]. GO has been applied in the clean-up of guttation liquid samples to analyze insecticides, providing negligible ME and analyte recoveries between 70% and 120% with relative standard deviations < 20% for most compounds tested [88]. Yadav et al. [89] described using GO to extract organophosphorus pesticides from water samples. Reduced graphene oxide–titanium dioxide (rGO-TiO₂) nanomaterial has been successfully applied as a dispersive solid phase extraction (d-SPE) adsorbent to extract, enrich, and purify aflatoxins in poultry feed [90].

Fe₃O₄MNPs generally yielded comparable recovery values to the GO clean-up option but slightly reduced ramipril recovery below the acceptable value. Significantly, the ME for some pharmaceuticals was not reduced to the desired degree with either Fe₃O₄MNPs or PSA/C18/MgSO₄ and was −25% and −23%, 32% and 38%, 25% and 22%, −24% and −21%, 23% and 24%, and 25% and 26%, respectively. Clean-up with classical sorbents—PSA/C18/MgSO₄ and PSA/GCB/MgSO₄—negatively affected the recoveries of some compounds, including ramipril and venlafaxine. Like GO, chitosan significantly reduced the ME and recovery values below the acceptable limit for almost half of the compounds (Figure 4). The application of PSA/C18/MgSO₄ and Fe₃O₄MNPs resulted in moderate signal enhancement for indapamide, levonorgestrel, and two metabolites of venlafaxine, while suppression was observed for doxycycline and ramipril.

3.3. Validation Study

In the conducted study, linearity, recovery, precision, limit of detection (LOD), limit of quantification (LOQ), and uncertainty (U) were assessed to validate the method. The validation was performed under optimized extraction and clean-up conditions using model wastewater.

The linearity of the calibration curves was assessed using calibration solutions at concentrations of 0.005, 0.01, 0.05, 0.1, 0.5, and 1.0 μg⋅mL⁻¹ in solvent and wastewater extract on three consecutive days using LC-MS/MS. The evaluation results show that all compounds’ coefficient of determination (R²) was 0.99 or higher.

The recovery was determined in five replicates at three fortification levels: 0.001, 0.01, and 0.1 μg⋅mL⁻¹. All compounds showed satisfactory recoveries in the 71–97% range at 0.001 μg⋅mL⁻¹, 70–98% at 0.01 μg⋅mL⁻¹, and 73–99% at 0.1 μg mL⁻¹. All pharmaceuticals yielded RSD not exceeding 16%. The RSD of twelve compounds at each level was less than 10%.

The limit of quantification (LOQ) was determined as the lowest enrichment level confirmed by satisfactory recovery (70–120%) and RSD (≤20%) values. The LOQ was in the range of 0.15–0.70 μg⋅mL⁻¹ for all compounds. The limit of detection (LOD) values of individual pharmaceuticals were calculated using the chromatogram when the signal-to-noise (S/N) ratio was higher than 3 [91].

The data obtained from the validation study were utilized to estimate the measurement uncertainty (U) associated with the analytical results. The expanded measurement uncertainties were determined using a “top-down” empirical model [92], with values ranging from 9% to 28% (coverage factor k = 2, confidence level 95%). Precision was identified as the primary source of uncertainty. Additionally, the uncertainty associated with recovery, calculated based on a rectangular distribution, was incorporated into the overall uncertainty budget to prevent an underestimation of the total uncertainty.

The results are presented in

Table 4. Average recoveries, LODs, LOQs, and expanded uncertainties U, % (k = 2, confidence level 95%) of optimized procedure.

Pharmaceutical or Metabolite	LOD (μg mL−1)	LOQ (μg mL−1)	Recovery (RSD) (%)			U (%)	ME (%)
			0.001 (μg⋅mL−1)	0.01 (μg⋅mL−1)	0.1 (μg⋅mL−1)
Atenolol	0.05	0.15	96 (8)	97 (4)	98 (4)	11	4
Bisoprolol	0.06	0.20	71 (8)	70 (4)	75 (6)	14	−2
Candesartan	0.20	0.65	85 (5)	86 (5)	84 (7)	11	4
Ciprofloxacin	0.09	0.30	81 (4)	80 (9)	79 (11)	13	8
Doxycycline	0.15	0.50	97 (16)	98 (5)	95 (5)	11	−10
Duloxetine	0.13	0.45	96 (12)	94 (13)	96 (12)	20	−1
Escitalopram	0.15	0.50	90 (6)	88 (9)	90 (9)	21	7
Fluoxetine	0.10	0.35	95 (6)	96 (8)	99 (4)	14	−2
Indapamide	0.15	0.50	75 (6)	73 (3)	75 (6)	11	12
Levonorgestrel	0.15	0.50	87 (6)	84 (6)	87 (9)	17	15
Metformin	0.10	0.35	85 (10)	84 (13)	82 (6)	26	8
Metoprolol	0.08	0.25	84 (7)	73 (9)	74 (6)	20	−9
Propafenone	0.08	0.25	82 (5)	80 (5)	82 (5)	8	−7
Ramipril	0.09	0.30	72 (7)	70 (3)	73 (5)	9	−11
Sertraline	0.13	0.45	72 (10)	74 (7)	76 (6)	15	7
Torasemide	0.19	0.65	90 (11)	92 (4)	94 (6)	14	2
Valsartan	0.15	0.50	97 (6)	95 (11)	93 (8)	22	−6
Venlafaxine	0.22	0.70	74 (9)	73 (5)	74 (6)	9	2
Venlafaxine N-oxide	0.20	0.65	73 (8)	71 (6)	73 (4)	20	12
Venlafaxine-O-desmethyl	0.07	0.25	83 (11)	82 (3)	84 (6)	7	11

, demonstrating the proposed method’s suitability.

4. Conclusions

A robust and sensitive analytical method for determining trace-level pharmaceuticals in wastewater was successfully developed and validated. The approach combines a modified QuEChERS extraction procedure with graphene oxide-based clean-up and LC-MS/MS detection, resulting in high recovery rates, acceptable precision, low limits of quantification, and minimal matrix effects. Incorporating GO significantly improved clean-up efficiency without compromising analyte recovery, enhancing the overall reliability of the method.

Due to its simplicity, low cost, and analytical performance, the method is suitable for routinely monitoring pharmaceutical residues in complex environmental matrices such as wastewater. Beyond wastewater, this strategy holds potential for adaptation to other sample types, such as surface waters, sediments, or sludge; nevertheless, it requires further development and validation. Further development may focus on expanding the scope to additional emerging contaminants, miniaturizing the sample preparation process, or integrating the method into on-site or automated monitoring systems. These improvements could expand its applicability in environmental surveillance and regulatory frameworks aimed at reducing pharmaceutical pollution.