DOI: https://doi.org/10.21203/rs.3.rs-2042203/v1
Perfluorinated compounds (PFCs) are a wide class of emerging pollutants still under study. In this work, we developed and validate a sensitive analytical method based on HPLC-MS/MS for the determination of 21 PFCs. This method was then used to investigate the presence of the target PFCs in six rivers in central Italy during a 4-months long monitoring campaign. 73% of the analytical determinations resulted higher than the limit of detection (LOD). The ∑21PFCs ranged from 4.3 to 68.5 ng L− 1 with the highest concentrations measured in June month, due to a minor river streamflow occurring in the warm periods. Between the individual congeners, PFBA and PFPeA, followed by PFHxA and PFOA were the predominant congeners detected. The evidence that short and medium chain PFCs (C4-C9) prevail over the long chain PFCs (C10-C18) could be attributed to the increased use and higher solubility of short chain PFCs compared to long chain PFCs. The ecological risk assessment, conducted by using risk quotient (RQ) method, highlighted that for PFBA, PFPeA, PFBS, PFHxA and PFOA the risk for aquatic environments was low or negligible. Only for PFOA there was a medium risk in 2 rivers in June month. As regard PFOS, 54% of the river water samples were classified as “high risk” for the aquatic environment. The remaining 46% of the samples were classified as “medium risk”.
Perfluorinated compounds (PFCs) are a wide class of organic substances characterized by a fluorinated hydrophobic carbon chain, generally bound to a hydrophilic head group (González-Barreiro et al. 2006; Kancharla et al. 2022). Because of their structure, PFCs present unique properties, such as surface activity, thermal and acid resistance and repellency of both water and oil, that make them ideal for several commercial and industrial applications (Wang et al. 2017a). Since 1960s, PFCs were used as constituents of a wide range of products including fluoropolymers (e.g. polytetrafluoroethylene; PTFE), liquid repellents for paper, food-packaging, cookware, textiles, leather, carpet, and firefighting foams (Wang et al. 2009; Blaine et al. 2013; Ojemaye and Petrik, 2019; Kurwadkar et al. 2022). Due to the high energy of carbon–fluorine (C–F) bonds, PFCs are extremely resistant to any degradation process, such as hydrolysis, photolysis, metabolism, and biodegradation (Li et al. 2012; Organisation for Economic Co-operation and Development (OECD), 2018). This characteristic, together with their water solubility, tendency to bioaccumulate and to biomagnify, has determined their ubiquitous distribution in the environment, wildlife, and humans across the world (Butt et al. 2010; Zhang et al. 2014; Campo et al. 2016; Lam et al. 2017; Sedlak et al. 2017; Boisvert et al. 2019). Among PFCs, perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) have been the most widely studied (Post et al. 2012; Li et al. 2013; Miralles-Marco and Harrad, 2015; Liang et al. 2022). In 2009, PFOS and its salts were included in the Stockholm Convention as new Persistent Organic Pollutants (new POPs) due to their persistency, ability to bioaccumulate and toxicological concern (UNEP, 2009; Domingo, 2012). As concerns PFOA, firstly its manufacture and use were being phased out to reduce global emissions (USEPA, 2010a, 2012) and in 2019 PFOA, its salts and PFOA-related compounds has been listed in Annex A of the Stockholm Convention (UNEP, 2019). However, due to resistance to degradation and chemical inertia characterizing this class of compounds, human exposure and environmental contamination are expected to continue for the near future and beyond (Lindrom et al. 2011a; Post et al. 2012; Ahmed et al. 2020). Recent researches indicate that the general population is exposed to PFCs through different pathways, such as inhalation, ingestion, and dermal absorption (Genuis et al. 2010; Stahl et al. 2011; Jurado-Sánchez et al. 2014; Ledda et al. 2018; Ahrens et al. 2019; Canova et al. 2020; Dalla Zuanna et al. 2021; Claessens et al. 2022). Some toxicological studies suggest that PFCs bind to serum proteins causing adverse effects on human health, such as the perturbation in the fatty acid metabolism, endocrine disruption, increase of blood cholesterol levels, and the induction of adverse effects in liver, kidneys, and other tissues (Anderson-Mahoney et al. 2008; Li et al. 2013; Jurado-Sánchez et al. 2014; ATSDR, 2018; Pelch et al. 2019). Moreover, several studies focused on how PFCs and other organic pollutants act synergistically in causing several health disorder (Hamid and Li, 2016; Jain, 2019; Garg et al. 2020). The biological half-life of PFCs in human body is quite long and depends both on the chemical-physical characteristics of the substance (e.g. carbon-chain length, functional group of the molecule), and on the characteristics of the subject exposed to PFCs (e.g. age and sex) (Li et al. 2017). For example, PFOS and PFOA have a median human serum half-life of 3.4 years and 2.7 years, respectively (Li et al. 2018).
Over the last years, given the human and ecological health concerns associated with this class of compounds, the scientific community has focused interest in PFCs determination, mainly in the water matrix (Thompson et al. 2011; Barreca et al. 2018; Deng et al. 2019; Zhang et al. 2019; Fauconier et al. 2020; Zhang et al. 2021). In Italy, the analytical determinations of these substances in water samples started in 2013, after the discovery of massive groundwater contamination in a vast area in northern Italy (Valsecchi et al. 2015; Bonelli et al. 2020; Chiesa et al. 2022). Although the nature of these pollutants does not exclude large-scale deployment, the determination of PFAs was carried out only in the surface waters of northern Italy.
In this study we develop and validate a sensitive analytical method for the determination of 21 PFCs (C4-C14, C16, and C18 for perfluoroalkyl carboxylic acids and C4-C10 and C12 for perfluoroalkyl sulfonic acids). The choice of these compounds was made because, since this is an exploratory study, it was important to investigate the presence of compounds having different carbon chain length. The validated method was then applied to investigate the presence of the target PFCs in river water samples collected in Umbria region (central Italy) during a four-month sampling campaign (March-June 2022). To our knowledge, this is the first work that investigate the presence of PFCs in Umbria region (central Italy). In addition, the monitoring campaign conducted over several months allowed to study the monthly trend of these pollutants.
Chemicals and Reagents
Methanol (MeOH) LC–MS grade was supplied by Merck (Darmstadt, Germany), and ultrapure water was obtained from a Milli-Q filter system (Millipore, Bedford, MA, USA). Ammonium acetate, HPLC grade, was from Merck (Darmstadt, Germany). Stock standards containing 2 µg mL−1 of perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluoroundecanoic acid (PFUdA), perfluorododecanoic acid (PFDoA), perfluorotridecanoic acid (PFTrDA), perfluorotetradecanoic acid (PFTeDA), perfluorohexadecanoic acid (PFHxDA), perfluorooctadecanoic acid (PFODA), potassium perfluoro-1-butanesulfonate (PFBS), sodium perfluoro-1-pentanesulfonate (PFPeS), sodium perfluoro-1-hexanesulfonate (PFHxS), sodium perfluoro-1-heptanesulfonate (PFHpS), sodium perfluoro-1-octanesulfonate (PFOS), sodium perfluoro-1-nonanesulfonate (PFNS), sodium perfluoro-1-decanesulfonate (PFDS), sodium perfluoro-1-dodecanesulfonate (PFDoS), were obtained from Wellington Laboratories Inc. (Guelph, Ontario, Canada). Mass-labelled injection standards (IS) at a concentration of 2 µg mL−1 of perfluoro-n-(2,3,4-13C3) butanoic acid (M3PFBA), perfluoro-n-(1,2-13C2) octanoic acid (M2PFOA), perfluoro-n-(1,2-13C2) decanoic acid (MPFDA) and sodium perfluoro-1-(1,2,3,4-13C4) octane sulfonate (MPFOS) were purchased from Wellington Laboratories Inc. (Guelph, Ontario, Canada). Mass-labelled extraction standards (ES) at a concentration of 2 µg mL-1 of perfluoro-n-(13C4)butanoic acid (MPFBA), perfluoro-n-(13C5)pentanoic acid (M5PFPeA), perfluoro-n-(1,2,3,4,6-13C5)hexanoic acid (M5PFHxA), perfluoro-n-(1,2,3,4-13C4)heptanoic acid (M4PFHpA), perfluoro-n-(13C8)octanoic acid (M8PFOA), perfluoro-n-(13C9)nonanoic acid (M9PFNA), perfluoro-n-(1,2,3,4,5,6-13C6)decanoic acid (M6PFDA), perfluoro-n-(1,2,3,4,5,6,7-13C7)undecanoic acid (M7PFUdA), perfluoro-n-(1,2-13C2)dodecanoic acid (MPFDoA), perfluoro-n-(1,2-13C2)tetradecanoic acid (M2PFTeDA), sodium perfluoro-1-(2,3,4-13C3)butanesulfonate (M3PFBS), sodium perfluoro-1-(1,2,3-13C3)hexanesulfonate (M3PFHxS), and sodium perfluoro-1-(13C8)octanesulfonate (M8PFOS) were purchased from Wellington Laboratories Inc. (Guelph, Ontario, Canada). Stock solutions (20 ng mL−1) of the analytes were prepared in MeOH:H2O (80:20, v/v), in polypropylene (PP) volumetric tubes, and then stored at 4 °C. Water–methanol (80:20, v/v) calibration solutions at concentrations 0.2, 1.0, 10 and 20 ng mL-1 were freshly prepared before each measurement and stored at +4°C. The long-term stability of stocks was monitored to guarantee the consistency of standards. For the ultimate method validation, a certified reference material (CRM) IRMM-428, purchase by the Institute for Reference Materials and Measurements (Belgium), was analyzed.
Samples Collection
During March to June, 2022, samples of surface water were collected from six different rivers in the central-northern area of Umbria region (central Italy; Figure 1). The choice of the most appropriate monitoring sites was made on the basis of a previous study conducted by ARPA Umbria (regional agency for environmental protection) (Nucci et al. 2019a and Nucci et al. 2019b; Charavgis et al. 2022). The results obtained from this previous investigation highlighted the necessity of monthly-monitoring of PFCs, both in rivers where the concentrations of these substances exceeded the maximum levels fixed by directive 2013/39/EU (European Commission, 2013) and in those affected by significant pressure factors (NES, CAI, GEN, TVN; Table 1). Details regarding sampling dates, locations and area description of the investigated rivers are given in Table 1. All the samples were stored in a 1 L polypropylene (PP) tube, pre-cleaned with methanol followed by ultrapure water in order to avoid contaminations. The samples were stored in a coolbag, transported to the laboratory and stored at +4°C until analysis.
Table 1. Sampling locations, date, area description and major type of contamination.
Site |
Sampling Dates |
Longitude |
Latitude |
Area description |
Type of contamination |
|
CAI_mar CAI_apr CAI_may CAI_jun |
Caina |
03/03/2022 04/05/2022 05/10/2022 06/03/2022 |
12°15'44.73"E |
43°0'9.06"N |
Rural Area |
Water treatment plants, urban waste water, and agro-livestock |
NES_mar NES_apr NES_may NES_jun |
Nestore |
03/03/2022 04/05/2022 05/10/2022 06/03/2022 |
12°21'58.61"E |
42°54'27.60"N |
Urban area/Rural Area |
Water treatment plants, urban waste water, and agro-livestock |
GEN_mar GEN_apr GEN_may GEN_jun |
Genna |
03/03/2022 04/05/2022 05/11/2022 06/04/2022 |
12°17'29.24"E |
42°58'8.93"N |
Rural Area |
Water treatment plants, urban waste water, and agro-livestock |
TOP_mar TOP_apr TOP_may TOP_jun |
Topino |
03/02/2022 04/05/2022 05/11/2022 06/04/2022 |
12°30'27.33"E |
43° 1'34.51"N |
Urban Area/Rural Area |
Industrial plants, water treatment plants, and urban waste water |
SAO_mar SAO_apr SAO_may SAO_jun |
Saonda |
03/07/2022 04/05/2022 05/12/2022 06/05/2022 |
12°39'25.20"E |
43°15'45.79"N |
Rural Area |
Industrial plants, urban waste water, and agro-livestock |
TVN_mar TVN_apr TVN_may TVN_jun |
Timia |
03/07/2022 04/05/2022 05/12/2022 06/05/2022 |
12°36'38.53"E |
42°55'50.37"N |
Urban area/Industrial Area |
Industrial plants, water treatment plants, and urban waste water |
Samples Extraction and Instrumental Analysis
An aliquot of 250 mL of the water sample was collected in a 250 mL high-density polypropylene (HDPE) bottle with a narrow neck. The aliquot was spiked with 250 µL of the ES at the concentration of 20 ng mL-1 and intensively mixed with a vortex mixer. The sample was than extracted via solid phase extraction (SPE) using Strata™-XL-AW cartridge (100 mg, 6 mL, Phenomenex, CA, United States). The cartridge was previously conditioned with 10 mL of MeOH followed by 10 mL of phosphate buffer 0.1 M. The sample was then passed through the cartridge with the aid of a membrane pump at a flow rate of 5 mL min-1 and cleaned with 10 mL of ammonium acetate (1 g L-1) followed by 1 mL of MeOH. Subsequently, the cartridge was dried for 5 min under high vacuum (15-20 mm Hg). The target compounds were than eluted with 2 × 5 mL MeOH:NH4OH and dried under a gentle flux of nitrogen (purity >99.999%). The sample was then re-suspended in 250 μL of IS (20 ng mL-1) and analyzed with the HPLC–MS/MS method. Blanks (Milli-Q H2O), fortified with ES solution at the same concentration of the samples, were prepared and analyzed simultaneously. Analysis were performed using HPLC Agilent 1290 Infinity II (Agilent Technologies, Santa Clara, CA), fitted with an auto-sampler (Agilent 1260 G7129A), and coupled by Jet Stream electrospray ionization (AJS-ESI) to a tandem mass spectrometry (ESI-triple quadrupole Agilent Jet Stream 6460) operating in MRM (Multiple Reaction Monitoring) mode. Analyses were achieved in negative mode. The chromatographic column was a Zorbax Eclipse Plus C18 RRHD (50 x 3.0 mm, 1.8 μm) purchased from Agilent Technologies (Santa Clara, CA). A delay column (Zorbax Eclipse Plus C18 RRHD, 4.6 x 30 mm, Agilent Technologies, USA) was installed between the solvent mixer and injector module to avoid instrumental contamination.
Validation, QA and QC
The method was validated by testing repeatability, reproducibility, specificity, sensitivity and recovery. For the quantification of the samples, the isotope dilution method was applied. Thirteen mass-labelled compounds (ES) were used as surrogate standards with the purpose to determine the relative response factor (RRF) of the related native compound and to verify the retention time (RT). For the native compounds without the corresponding ES, the one with the most suitable structure was chosen (Table 2). Repeatability and reproducibility were evaluated on seven independent tests, by spiking different aliquots of the same river water sample at different concentrations (0.2, 0.65, 5, 10 and 20 ng L−1). All the tests were performed in sevenfold. Five points calibration curve covering concentrations from 0.2 ng mL−1 to 20 ng mL−1 was used for the quantification of PFASs in environmental samples. The limits of quantification (LOQs) of each analyte were defined as the concentration value equal or below a value of 30% of the relevant environmental quality standards (EQS) as established in commission directive 2009/90/EC. As concern the method recoveries, firstly these were assessed by analyzing water samples artificially contaminated with target analytes at different concentrations (0.2, 0.65, 5, 10 and 20 ng L-1). Even in this case, all the tests were performed in sevenfold. Moreover, recoveries were assessed by spiking every samples with ES at the concentration of 20 ng mL-1. Quality-control (QC) standards (one procedural blank sample and one calibration standard at the concentration of 1 ng mL−1) were analyzed every tenth sample, in order to control the instrument sensitivity. The ion ratio between the qualitative MS/MS transition response relative to the quantitative MS/MS transition response and the RT were recorded for each analyte and every sample, in order to correctly identify the compounds. The performance of the instrument was also monitored by adding to the samples IS at 20 ng mL-1 just before the injection. The method reliability was also examined by analyzing the certified reference material IRMM-428 (tap drinking water).
Multivariate statistical analysis
Multivariate statistical analysis was carried out by using the statistical software R (R-project for statistical computing, Ver. 3.0, 32-bit). Principal component analysis (PCA) was performed on the medium concentrations of PFCs in order to cluster the tracers of the main emission sources. Before performing the PCA, the matrix of the data was transformed by column mean centering and row and column autoscaling to correct for different variable scaling and units.
Chromatography was performed using H2O (A) – MeOH (B) both containing 2 mM of ammonium acetate at a flow rate of 500 µL min− 1. Gradient elution started at 40% of B for 0.5 minutes and was raised to 80% within 7.5 minutes; after 4 minutes in isocratic condition, mobile phase B was raised to 95%, and equilibrated for 1 minutes. The initial conditions were then restored and the system was equilibrated for 2 minutes. The column temperature was set at 40°C, and the injection volume was 5 µL. The RTs of the analytes were listed in Table 2. AJS-ESI-MS/MS measurements were performed in negative MRM detection mode. Mass spectrometer parameters were obtained by tuning the electrical parameters for each compound by infusion of standard solutions at concentration of 1 µg mL− 1 at flow rate of 0.7 µL min− 1. The source gas temperature and the sheath gas temperature were set at 320°C and 350°C, respectively. The ion capillary (IS) and the nozzle voltage were 3750 V and 1500 V, respectively. The gas flow and the sheath gas flow were set at 5 L min− 1 and 12 L min− 1. The nebulizer was 50 psi and the cell accelerator voltage was 7 V for all the analytes under study. The optimized parameters for each analyte are reported in Table 2, together with precursor and product ions and the mass labeled compound chosen as internal standard. Figure 2 shows a HPLC-MS/MS chromatogram obtained by injecting 5 µL of the PFCs standard solution at 20 µg L− 1.
Internal Standard | RT (min) | Precursor Ion (m/z) | Product Ion (m/z) | Fragmentor (V) | Collision Energy (eV) | |||
---|---|---|---|---|---|---|---|---|
PFBA | M3PFBA | 0,81 | 213 | 169 | 60 | 8 | ||
PFPeA | M5PFPeA | 1,98 | 263 | 219 | 60 | 6 | ||
PFBS | M3PFBS | 2,08 | 298,9 | Q | 80 | 133 | 45 | |
q | 98,9 | 29 | ||||||
PFHxA | M5PFHxA | 3,56 | 312,9 | 268,9 | 66 | 5 | ||
PFPeS | M3PFHxS | 3,82 | 349 | Q | 80 | 135 | 40 | |
q | 99 | 36 | ||||||
PFHpA | M4PFHpA | 5 | 362,9 | Q | 319 | 66 | 5 | |
q | 169 | 13 | ||||||
PFHxS | M3PFHxS | 5,13 | 398,9 | Q | 80 | 174 | 49 | |
q | 99 | 45 | ||||||
PFOA | M8PFOA | 6,1 | 412,9 | Q | 368,9 | 86 | 5 | |
q | 169 | 13 | ||||||
PFHpS | M8PFOS | 6,17 | 449 | Q | 80 | 100 | 50 | |
q | 99 | 46 | ||||||
PFNA | M9PFNA | 7,09 | 462,9 | Q | 418,9 | 66 | 5 | |
q | 169 | 17 | ||||||
PFOS | M8PFOS | 7,09 | 498,9 | Q | 80 | 210 | 50 | |
q | 99 | 50 | ||||||
PFDA | M6PFDA | 7,74 | 512,9 | Q | 469 | 102 | 5 | |
q | 169 | 20 | ||||||
PFNS | M8PFOS | 7,65 | 549 | Q | 80 | 165 | 76 | |
q | 99 | 48 | ||||||
PFUdA | M7PFUdA | 8,45 | 562,9 | Q | 519 | 92 | 5 | |
q | 169 | 21 | ||||||
PFDS | M8PFOS | 8,38 | 598,9 | Q | 80 | 120 | 94 | |
q | 99 | 60 | ||||||
PFDoA | MPFDoA | 9,04 | 612,9 | Q | 569 | 97 | 5 | |
q | 169 | 25 | ||||||
PFTriA | M2PFTeDA | 9,56 | 662,9 | Q | 619 | 102 | 9 | |
q | 169 | 30 | ||||||
PFDoS | M8PFOS | 9,42 | 699 | Q | 80 | 100 | 64 | |
q | 99 | 60 | ||||||
PFTeA | M2PFTeDA | 10,08 | 712,9 | Q | 669 | 112 | 9 | |
q | 169 | 40 | ||||||
PFHxDA | M2PFTeDA | 11,92 | 813 | Q | 769 | 100 | 15 | |
q | 169 | 40 | ||||||
PFODA | M2PFTeDA | 13,38 | 913 | Q | 869 | 200 | 15 | |
q | 169 | 40 | ||||||
Mass-labelled extraction standards (ES) | MPFBA | 0,89 | 217 | 172 | 60 | 8 | ||
M5PFPeA | 1,85 | 268 | 223 | 60 | 6 | |||
M3PFBS | 2,13 | 302 | 80 | 133 | 45 | |||
M5PFHxA | 3,43 | 318 | 273 | 66 | 5 | |||
M4PFHpA | 4,83 | 367 | 322 | 66 | 5 | |||
M3PFHxS | 4,96 | 402 | 80 | 174 | 49 | |||
M8PFOA | 5,96 | 421 | 376 | 86 | 5 | |||
M9PFNA | 6,87 | 472 | 427 | 66 | 5 | |||
M8PFOS | 6,9 | 507 | 80 | 210 | 50 | |||
M6PFDA | 7,64 | 518,9 | 473,9 | 102 | 5 | |||
M7PFUdA | 8,3 | 570 | 525 | 92 | 5 | |||
MPFDoA | 8,86 | 614,9 | 570 | 97 | 5 | |||
M2PFTeDA | 9,92 | 715 | 670 | 112 | 5 | |||
Mass-labelled injection standards (IS) | M3PFBA | 0,9 | 216 | 172 | 60 | 8 | ||
M2PFOA | 5,96 | 415 | 370 | 86 | 5 | |||
MPFOS | 6,9 | 503 | 99 | 210 | 50 | |||
MPFDA | 7,65 | 515 | 470 | 102 | 5 |
To verify the accuracy of the presented method in quantifying very low concentrations expected in superficial water samples, a validation experiment is performed; briefly, different aliquots of the same water sample were spiked at different concentrations (0.2, 0.65, 2, 5, 10 and 20 ng L− 1), extracted and injected. The tests were performed in sevenfold. The different concentration levels were chosen as follow: 0.2 ng L− 1 and 0.65 ng L− 1 represent respectively LOQ and environmental quality standard (EQS) for PFOS, and the concentrations between 5 and 20 ng L− 1 represent PFCs environmental contamination levels, determined in a previous investigation (Nucci et al. 2019a and Nucci et al. 2019b). The results obtained are shown in Table 3. In accordance with EPA (Environmental Protection Agency) method 533 (2021), for analytes fortified at concentration ≤ 2 times LOQ level, the results with mean recovery ranging from 50 to 150% were acceptable. For analyte fortified at concentration > 2 times LOQ level, the acceptable recovery range is within 70–130% of the true value. All the obtained values, for the three concentration levels, are in line with the acceptance criteria set out in EPA method 533, with mean percent recoveries ranging from 51–133% for low level, from 84–129% for medium level, and from 84–113% for high level. The only exception is represented by PFODA; as can be seen from Table 3, for high level the mean percent recoveries is 57%. For this reason, this compound is only qualified in this study. The obtained recoveries were comparable or higher than those obtained by Juricova et al. (2022). The intraday relative standard deviation (RSD%) for low level ranged from 2–16%, from 0.2–25% for medium level and from 0.3–8% for high level. The R2 was greater than 0.999 for all the analytes under study, with the exception of PFHpA, PFOA, PFHpS and PFDS (Table 3). During the method development and in every batch of river water samples, investigation of blank samples was also carried out to monitor background contamination. All the PFCs detected in the blank samples showed concentration levels less than 1/3 of the corresponding LOQ, as indicated in the EPA method 533.
Low level | Medium level | High level | ||||||
---|---|---|---|---|---|---|---|---|
LOQ | R2 | Mean %R | RSD% | Mean %R | RSD% | Mean %R | RSD% | |
PFBA | 5.0 | 0.999 | 100 | 3 | 84 | 0.2 | 84 | 0.3 |
PFPeA | 0.65 | 0.999 | 94 | 11 | 117 | 7 | 90 | 2 |
PFBS | 0.65 | 0.999 | 107 | 6 | 111 | 6 | 96 | 5 |
PFHxA | 0.65 | 0.999 | 124 | 9 | 126 | 5 | 103 | 5 |
PFPeS | 0.65 | 0.999 | 88 | 4 | 110 | 7 | 87 | 5 |
PFHpA | 0.65 | 0.997 | 114 | 5 | 123 | 9 | 99 | 4 |
PFHxS | 0.65 | 0.999 | 104 | 5 | 115 | 8 | 102 | 4 |
PFOA | 5.0 | 0.997 | 114 | 3 | 86 | 2 | 92 | 3 |
PFHpS | 0.65 | 0.998 | 93 | 7 | 113 | 8 | 93 | 3 |
PFNA | 0.65 | 0.999 | 102 | 2 | 121 | 8 | 98 | 3 |
PFOS | 0.20 | 0.999 | 133 | 2 | 104 | 5 | 113 | 7 |
PFNS | 0.65 | 0.999 | 94 | 3 | 100 | 25 | 93 | 4 |
PFDA | 0.65 | 0.999 | 95 | 2 | 114 | 6 | 89 | 2 |
PFDS | 0.65 | 0.998 | 100 | 7 | 95 | 11 | 108 | 8 |
PFUdA | 0.65 | 0.999 | 99 | 7 | 124 | 9 | 96 | 4 |
PFDoA | 0.65 | 0.999 | 89 | 5 | 118 | 7 | 90 | 2 |
PFDoS | 0.65 | 0.999 | 84 | 16 | 93 | 7 | 85 | 7 |
PFTrDA | 0.65 | 0.999 | 103 | 9 | 129 | 9 | 112 | 8 |
PFTeA | 0.65 | 0.999 | 94 | 5 | 122 | 5 | 103 | 5 |
PFHxDA | 0.65 | 0.999 | 115 | 7 | 127 | 12 | 109 | 6 |
PFODA | 0.65 | 0.999 | 51 | 6 | 99 | 10 | 57 | 11 |
The reliability of the proposed method has been also verified by analysis of the reference material IRMM-428 containing seven analytes (PFBS, PFHxS, PFOS, PFPeA, PFHxA, PFHpA and PFNA). The certified concentrations ranged from 3.6 to 9.6 ng L− 1; the comparison between measured and certified values is presented in Fig. 3.
The good results achieved in the validation experiments and in the analysis of the certified material make this method suitable for the analysis of PFCs in water samples.
The results of the analysis of river water samples (n = 24) are summarized in Table 4. Target PFCs were detected in all the analyzed river water samples, but the levels of the analytes vary widely between months and sampling points. The ∑21PFCs in river water samples ranged from 4.3 to 68.5 ng L− 1 (Table 4). All the analyzed river water samples were in line with the EQS established by Italian 172/2015 Decree Law (1000 ng L− 1 for PFHxA, 3000 ng L− 1 for PFPeA and PFBS, 7000 ng L− 1 for PFBA, and 100 ng L− 1 for PFOA). The only exception was represented by PFOS (EQS: 0.65 ng L− 1); as can be seen in Fig. 3, 46% of the analyzed samples exceed the fixed EQS for PFOS (Fig. 4, red line). For all the considered rivers, PFOS concentrations recorded in June were higher than the fixed EQS. The only exception was represented by TOP river, in which PFOS did not exceed the established EQS in any sampling month. On the contrary, GEN was the only river in which PFOS exceeded the fixed EQS in every sampling month (Fig. 4). Pignotti et al. (2017) and Zhu et al. (2015) reported a seasonal trend in which PFOS maximum concentration was recorded in winter and spring, respectively.
PFOS concentration range found in this study varied between < LOD and 2.0 ng L− 1. These values were in accordance with PFOS concentrations recorded by Yamazaki et al. (2016) and Yang (2016) in Chinese river water samples and by Ahrens et al (2009) in German rivers, but much lower than those reported by Navarro et al. (2020) in Spanish river waters. Pignotti and Dinelli (2018) studied the distribution of PFOS in several rivers in north Italy, found values comparable or higher than those reported in this study. Analyzing the individual PFCs contribution percentage, it appears that the short and medium chain PFCs (C4-C9) prevail over the long chain PFCs (C10-C18). This is in accordance with the trend found by Selvaraj et al. (2021) in Indian river waters and could be attributed to the increased use and higher solubility of short chain PFCs compared to long chain PFCs. The predominant congeners detected in this study were PFBA and PFPeA, followed by PFHxA and PFOA (Table 4). This trend is in accordance with that reported by Zhu et al. (2015). Navarro et al. (2020), instead, reported PFOS as predominant compound, followed by PFOA, PFHxA and PFHxS. The ∑21PFCs in water samples of March, April, May, and June ranged from 4.3 to 19.4 ng L− 1, from 9.7 to 36.6 ng L− 1, from 10.0 to 40.6 ng L− 1, and from 19.0 to 68.5 ng L− 1, respectively (Table 4 and Fig. 5). As shown in Fig. 4, the ÎŁ21PFCs was much higher in June for all the studied rivers, except for TOP river in which the ÎŁ21PFCs in May and June was comparable. The minimum ÎŁ21PFCs found in this study (4.3 ng L− 1) were comparable with that found by Zhu et al. (2015) in a Chinese river contaminated by several industrial waste but higher than that found by Navarro et al. (2020) in Spanish rivers. The maximum ÎŁ21PFCs found in this study (47.3 ng L− 1), instead, were much lower of those found by the same authors (Zhu et al. 2015), but higher than that found by Navarro et al. (2020) in Spanish rivers. Castiglioni et al. (2015) investigated the presence of PFCs in river water samples in north Italy, found values 19 times higher than those reported in this study.
PFBA | PFPeA | PFBS | PFHxA | PFPeS | PFHpA | PFHxS | PFOA | PFHpS | PFNA | PFOS | PFNS | PFDA | PFDS | PFUdA | PFDoA | PFDoS | PFTrDA | PFTeA | PFHxDA | PFODA | ∑21PFCs | ||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
LOD | 0.48 | 0.05 | 0.04 | 0.06 | 0.002 | 0.03 | 0.02 | 0.11 | 0.005 | 0.04 | 0.03 | 0.04 | 0.05 | 0.05 | 0.07 | 0.09 | 0.05 | 0.07 | 0.06 | 0.21 | 0.20 | ||
CAI | March | 2.2 | 2.5 | 0.5 | 2.2 | 0.1 | 0.8 | 0.4 | 2.0 | 0.04 | 0.4 | 0.8 | 0.1 | 0.4 | <LOD | 0.2 | 0.1 | <LOD | 0.1 | 0.1 | <LOD | <LOD | 12.9 |
April | 4.7 | 1.9 | 0.7 | 2.1 | 0.1 | 1.0 | 0.3 | 4.0 | 0.02 | 0.4 | 0.6 | 0.05 | 0.3 | <LOD | 0.2 | 0.1 | <LOD | 0.2 | 0.1 | <LOD | <LOD | 16.6 | |
May | 5.5 | 3.4 | 0.9 | 2.7 | <LOD | 1.2 | 0.04 | 1.8 | 0.6 | <LOD | 0.3 | 0.3 | 0.4 | <LOD | <LOD | <LOD | <LOD | <LOD | 0.2 | <LOD | <LOD | 17.2 | |
June | 11.6 | 12.3 | 3.9 | 10.2 | 0.2 | 3.5 | 1.7 | 7.9 | 0.1 | 1.3 | 2.0 | nd | 1.2 | nd | 0.1 | 0.1 | nd | 0.1 | < LOD | <LOD | <LOD | 56.2 | |
GEN | March | 3.0 | 4.4 | 0.7 | 3.9 | 0.1 | 1.4 | 0.8 | 2.7 | 0.04 | 0.4 | 1.0 | <LOD | 0.5 | nd | 0.1 | 0.1 | <LOD | 0.1 | 0.1 | <LOD | <LOD | 19.4 |
April | 6.1 | 7.5 | 1.5 | 6.5 | 0.2 | 3.1 | 0.7 | 7.1 | 0.1 | 0.9 | 1.8 | <LOD | 0.7 | <LOD | 0.2 | 0.1 | <LOD | 0.1 | 0.1 | <LOD | <LOD | 36.6 | |
May | 11.0 | 6.5 | 1.5 | 5.6 | 0.02 | 2.1 | 0.8 | 4.0 | 0.6 | 0.1 | 1.1 | 0.3 | 0.6 | <LOD | <LOD | <LOD | <LOD | <LOD | 0.2 | <LOD | <LOD | 34.6 | |
June | 14.5 | 12.3 | 2.6 | 12.4 | nd | 4.1 | 2.2 | 10 | nd | nd | 1.5 | nd | nd | nd | nd | 3.3 | nd | nd | nd | nd | nd | 63.0 | |
NES | March | 2.0 | 1.7 | 0.4 | 1.7 | 0.1 | 0.6 | 0.3 | 1.6 | 0.03 | 0.3 | 0.6 | <LOD | 0.3 | 0.3 | 0.1 | 0.1 | 0.05 | 0.1 | < LOD | <LOD | <LOD | 10.0 |
April | 4.5 | 1.2 | 0.7 | 1.7 | 0.03 | 1.0 | 0.2 | 3.8 | 0.02 | 0.4 | 0.7 | <LOD | 0.4 | <LOD | 0.1 | 0.1 | <LOD | 0.1 | 0.1 | <LOD | 0.2 | 15.3 | |
May | 14.7 | 7.9 | 1.4 | 5.2 | <LOD | 2.3 | 0.6 | 4.9 | 0.8 | 0.4 | 1.3 | <LOD | 0.9 | <LOD | <LOD | <LOD | <LOD | <LOD | 0.3 | <LOD | <LOD | 40.6 | |
June | 19.2 | 10.5 | 2.3 | 12.4 | 0.2 | 6.5 | 1.1 | 11 | 0.1 | 2.2 | 1.8 | nd | 1.1 | nd | 0.1 | 0.1 | nd | 0.1 | < LOD | <LOD | nd | 68.5 | |
SAO | March | 2.3 | 0.8 | 0.9 | 1.1 | 0.02 | 0.3 | 0.1 | 1.3 | 0.01 | 0.2 | 0.2 | 0.1 | 0.1 | 0.0 | 0.1 | 0.1 | <LOD | 0.1 | 0.1 | <LOD | <LOD | 7.8 |
April | 1.8 | 0.9 | 0.8 | 1.1 | <LOD | 0.4 | 0.1 | 3.5 | 0.01 | 0.2 | 0.2 | <LOD | 0.1 | <LOD | 0.1 | <LOD | 0.1 | 0.1 | 0.1 | <LOD | 0.2 | 9.7 | |
May | 3.6 | 1.7 | 1.7 | 1.1 | <LOD | 0.5 | < LOD | 0.4 | 0.6 | <LOD | <LOD | 0.3 | 0.0 | <LOD | <LOD | <LOD | <LOD | <LOD | 0.3 | < LOD | <LOD | 10.0 | |
June | 9.7 | 4.6 | 4.5 | 4.8 | 0.1 | 1.4 | 0.3 | 5.0 | 0.05 | 0.7 | 0.8 | nd | 0.7 | nd | 0.1 | 0.1 | nd | 0.1 | 0.1 | < LOD | <LOD | 33.0 | |
TOP | March | 1.3 | 0.4 | 0.2 | 0.6 | nd | 0.2 | 0.1 | 0.8 | 0.01 | 0.1 | 0.1 | 0.04 | 0.1 | 0.1 | 0.1 | 0.1 | <LOD | 0.1 | 0.1 | < LOD | <LOD | 4.3 |
April | 2.2 | <LOD | 0.3 | 1.1 | <LOD | 0.4 | 0.1 | 3.3 | 0.1 | 0.3 | 0.5 | 0.1 | 0.2 | 0.1 | 0.3 | 0.2 | 0.1 | 0.2 | 0.2 | 0.3 | 0.2 | 10.3 | |
May | 10.3 | 4.2 | 0.3 | 1.6 | <LOD | 0.5 | <LOD | 1.8 | 0.6 | <LOD | 0.1 | 0.3 | 0.2 | <LOD | 0.1 | <LOD | <LOD | 0.1 | 0.3 | <LOD | <LOD | 20.4 | |
June | 8.6 | 2.3 | 0.8 | 2.1 | nd | 0.6 | 0.1 | 3.1 | 0.04 | 0.3 | 0.5 | nd | 0.4 | nd | 0.1 | 0.1 | nd | <LOD | 0.1 | <LOD | nd | 19.0 | |
TVN | March | 4.3 | 1.5 | 0.6 | 2.4 | 0.1 | 0.5 | 0.2 | 2.3 | 0.03 | 0.4 | 0.5 | 0.1 | 0.3 | 0.1 | 0.2 | 0.1 | 0.05 | 0.1 | 0.1 | <LOD | <LOD | 13.7 |
April | 4.0 | 2.0 | 0.8 | 2.5 | 0.1 | 0.6 | 0.1 | 5.4 | 0.02 | 0.4 | 0.4 | 0.1 | 0.2 | <LOD | 0.1 | 0.1 | <LOD | 0.1 | 0.1 | 0.2 | <LOD | 17.2 | |
May | 6.3 | 3.3 | 0.7 | 2.4 | <LOD | 0.8 | <LOD | 1.1 | 0.6 | 0.0 | <LOD | 0.3 | 0.2 | <LOD | <LOD | <LOD | <LOD | <LOD | 0.3 | <LOD | <LOD | 15.9 | |
June | 6.9 | 5.1 | 1.3 | 3.6 | 0.05 | 0.8 | 0.2 | 3.1 | 0.03 | 0.4 | 0.7 | nd | 0.4 | nd | <LOD | 0.1 | nd | 0.1 | <LOD | <LOD | nd | 22.7 |
In this work, the risk assessment for the aquatic biota relating to PFCs presence and levels in surficial freshwaters of Umbria region was conducted by using risk quotient (RQ) method, according to Lv et al. (2019). Briefly, RQ is determined throughout the ratio between the measured environmental concentration (ng L− 1) and the corresponding EQS (ng L− 1). When the RQ value is ≥ 1, it indicates that the risk of contamination in the area is high. When 0.1 ≤ RQ < 1 and 0.01 ≤ RQ < 0.1 it means that there is, respectively, medium and low risk of contamination in the aquatic environment (Yan et al. 2013). The RQ values, calculated for 6 PFCs (PFBA, PFPeA, PFBS, PFHxA, PFOA and PFOS) detected in river water samples of Umbria region are reported in Table 5. The RQ values were calculated only for 6 pollutants because the Italian 172/2015 Decree Law fixed the EQS only for these compounds (1000 ng L− 1 for PFHxA, 3000 ng L− 1 for PFPeA and PFBS, 7000 ng L− 1 for PFBA, and 100 ng L− 1 for PFOA). For PFBA, PFPeA and PFBS the calculated RQ were much lower than 0.01 for all the investigated rivers and in all the sampling months (Table 5). These values indicate a negligible risk for the aquatic organisms. For PFHxA, the calculated RQ is between 0.01 and 0.1 (low risk for the aquatic ecosystem) for three rivers (CAI, GEN and NES) in June month (Table 5). Even if PFBA, PFPeA and PFHxA were among the most abundant pollutants detected in river waters (Table 4), their lower ability to bioaccumulate (in comparison to other monitored compounds) leads them to have rather high EQS (values between 1000 and 7000 ng L− 1; Valsecchi et al. 2017) and, consequently, low RQ values. As regard PFOA, the calculated RQ highlighted a low risk for the aquatic organisms for all the considered rivers in all sampling months. The only exceptions were NES and CAI rivers in June month, where the calculated RQ is higher than 0.1 (medium risk). As concern PFOS, the RQ values indicates a high risk for the aquatic ecosystem in 54% of the analyzed samples (Fig. 6). In the remaining 46% of the samples, the risk for the aquatic environment was classified as medium (Fig. 6).
PFBA | PFPeA | PFBS | PFHxA | PFOA | PFOS | ||
---|---|---|---|---|---|---|---|
EQS | 7000 | 3000 | 3000 | 1000 | 100 | 0.65 | |
CAI | March | 0.0003 | 0.0008 | 0.0002 | 0.0022 | 0.02 | 1.2 |
April | 0.0007 | 0.0006 | 0.0002 | 0.0021 | 0.04 | 1.0 | |
May | 0.0008 | 0.0011 | 0.0003 | 0.0027 | 0.02 | 0.4 | |
June | 0.0017 | 0.0041 | 0.0013 | 0.0102 | 0.08 | 3.1 | |
GEN | March | 0.0004 | 0.0015 | 0.0002 | 0.0039 | 0.03 | 1.6 |
April | 0.0009 | 0.0025 | 0.0005 | 0.0065 | 0.07 | 2.8 | |
May | 0.0016 | 0.0022 | 0.0005 | 0.0056 | 0.04 | 1.7 | |
June | 0.0021 | 0.0041 | 0.0009 | 0.0124 | 0.10 | 2.3 | |
NES | March | 0.0003 | 0.0006 | 0.0001 | 0.0017 | 0.02 | 0.9 |
April | 0.0006 | 0.0004 | 0.0002 | 0.0017 | 0.04 | 1.0 | |
May | 0.0021 | 0.0026 | 0.0005 | 0.0052 | 0.05 | 2.0 | |
June | 0.0027 | 0.0035 | 0.0008 | 0.0124 | 0.11 | 2.7 | |
SAO | March | 0.0003 | 0.0003 | 0.0003 | 0.0011 | 0.01 | 0.2 |
April | 0.0003 | 0.0003 | 0.0003 | 0.0011 | 0.04 | 0.3 | |
May | 0.0005 | 0.0006 | 0.0006 | 0.0011 | 0.004 | - | |
June | 0.0014 | 0.0015 | 0.0015 | 0.0048 | 0.05 | 1.2 | |
TOP | March | 0.0002 | 0.0001 | 0.0001 | 0.0006 | 0.01 | 0.2 |
April | 0.0003 | - | 0.0001 | 0.0011 | 0.03 | 0.7 | |
May | 0.0015 | 0.0014 | 0.0001 | 0.0016 | 0.02 | 0.2 | |
June | 0.0012 | 0.0008 | 0.0003 | 0.0021 | 0.03 | 0.8 | |
TVN | March | 0.0006 | 0.0005 | 0.0002 | 0.0024 | 0.02 | 0.7 |
April | 0.0006 | 0.0007 | 0.0003 | 0.0025 | 0.05 | 0.6 | |
May | 0.0009 | 0.0011 | 0.0002 | 0.0024 | 0.01 | - | |
June | 0.0010 | 0.0017 | 0.0004 | 0.0036 | 0.03 | 1.1 |
In the PCA, PFODA, PFHxDA, PFDoS and PFDS were excluded from the data analysis due to their low detection frequencies. PCA results are summarized in the biplot reported in Fig. 7, while scores and loadings are shown in Tables 6 and 7, respectively. Two significant components (PC1 and PC2), accounting for 85% of the total variance, were obtained. The biplot well separated two cluster of river water samples, each characterized by its emission profile (Fig. 7). The first cluster, in the left part of the biplot, consists of three river water samples (TVN, SAO and TOP) and four PFCs (PFTrDA, PFUdA, PFTeA and PFNS). All the previously mentioned rivers were affected by several industrial emission sources (paper industry, cement plants and other smaller industry). Unfortunately, a comparison with the literature is difficult due to the lack of data; to our knowledge no study investigated the release of PFCs from cement plants. Kim et al. (2012) analyzed wastewater treatment plants from different industrial activity, including paper mill, founding a contamination profile different form that reported in this study and dominated by C6-C8 congeners. The second cluster, in the right part of the biblot, consists of three river water samples (NES, CAI and GEN) and several PFCs (Fig. 7). All the three river waters composing this cluster were affected by discharging of urban wastewater and ago-livestock farms. Tuan et al. (2021) analyzed water samples collected in rivers affected by agricultural production, animal husbandry and discharge of urban wastewaters found high concentrations of short chain PFCs (PFBA, PFPeA, PFHxS, PFHxA).
PC1 | PC2 | PC3 | PC4 | PC5 | |
---|---|---|---|---|---|
% Variance | 67.7 | 16.7 | 7.4 | 6.7 | 1.4 |
CAI | 0.642971 | 0.668479 | 1.167002 | -1.80236 | -0.09673 |
GEN | 4.287088 | -2.48593 | 0.376748 | 0.494018 | 0.131172 |
NES | 3.56243 | 2.575559 | -0.70277 | 0.606793 | 0.065957 |
SAO | -2.48583 | -0.77456 | -1.65481 | -0.74759 | 0.427798 |
TOP | -3.96911 | 0.43549 | 1.199683 | 1.04909 | 0.387212 |
TVN | -2.03755 | -0.41904 | -0.38585 | 0.400051 | -0.91541 |
PC1 | PC2 | PC3 | PC4 | PC5 | |
---|---|---|---|---|---|
PFBA | 0.259044 | 0.169792 | 0.043688 | 0.350515 | 0.120222 |
PFPeA | 0.280738 | -0.12418 | 0.185783 | -0.04021 | -0.11494 |
PFBS | 0.117781 | -0.18936 | -0.46494 | -0.59145 | 0.5307 |
PFHxA | 0.287616 | -0.11215 | 0.087088 | -0.00112 | -0.09474 |
PFPeS | 0.25872 | 0.036042 | 0.089875 | -0.34861 | -0.56732 |
PFHpA | 0.292787 | 0.041764 | 0.037338 | 0.036653 | 0.140756 |
PFHxS | 0.266483 | -0.19057 | 0.224724 | -0.09228 | 0.150612 |
PFOA | 0.292666 | -0.03686 | 0.037423 | 0.084786 | 0.005009 |
PFHpS | 0.183203 | 0.425629 | 0.131584 | 0.200177 | 0.358498 |
PFNA | 0.20986 | 0.398796 | -0.12596 | -0.12943 | -0.07554 |
PFOS | 0.287013 | -0.03591 | 0.191045 | -0.02213 | 0.069016 |
PFNS | -0.18907 | -0.34496 | 0.411823 | -0.17455 | -0.05477 |
PFDA | 0.239623 | 0.290489 | 0.139629 | -0.25512 | 0.027632 |
PFUdA | -0.16936 | 0.168885 | 0.583882 | -0.35371 | 0.250011 |
PFDoA | 0.176316 | -0.43078 | 0.178784 | 0.220864 | 0.275016 |
PFTrDA | -0.23668 | 0.319927 | 0.143923 | -0.18057 | 0.045309 |
PFTeA | -0.28147 | 0.07269 | 0.161495 | 0.166624 | 0.185476 |
A sensitive analytical method based on offline SPE of 250 mL of river water sample and subsequent analysis by HPLC-MS/MS was validated for 21 PFCs. The method performances, in term of recovery, precision and sensibility, were satisfactory and in line with those established by EPA (EPA Method 533). The LOQs ranged from 0.2 to 5 ng L− 1 and allowed the detection of 21 PFCs in river water samples collected in six rivers of Umbria region (central Italy), for 4 consecutive months. As regard the PFCs levels recovered in all samples, concentrations were in line with the EQS established by Italian 172/2015 Decree Law, with the only exception of PFOS: 46% of the analyzed samples exceed the fixed EQS for this compound. As concern the congeners distribution, PFBA and PFPeA followed by PFHxA and PFOA were the predominant compounds. The study of the monthly distribution of these pollutants has highlighted that the ∑21PFCS was lower in march and grew towards the summer months for all the investigated rivers. The ecological risk assessment, based on the calculation of RQ, highlighted that for PFBA, PFPeA, PFBS, PFHxA and PFOA the risk for aquatic environments for all the rivers under study was low or negligible. Only for PFOA there was a medium risk for NES e GEN rivers in June month. As regard PFOS, there was a high risk for the aquatic environment in 54% of the river water samples, with RQ values between 1.0 and 3.1. For the remaining 46% of the samples the risk was medium, with RQ values between 0.2 and 0.96.
Funding
The authors declare that no funds, grants, or other support were received during the preparation of this manuscript
Competing Interests
The authors have no relevant financial or non-financial interests to disclose
Author Contributions
Conceptualization by Federica Castellani and Matteo Vitali; methodology by Mara Galletti, Fedra Charavgis, Alessandra Cingolani, Sonia Renzi, and Mirko Nucci; software by Federica Castellani; validation by Matteo Vitali, Mara Galletti, and Carmela Protano; formal analysis by Federica Castellani, Fedra Charavgis, Alessandra Cingolani, Sonia Renzi, and Mirko Nucci; investigation by Federica Castellani, Mara Galletti, Fedra Charavgis, Alessandra Cingolani, Sonia Renzi, Mirko Nucci, and Matteo Vitali; writing: original draft preparation by Federica Castellani and Matteo Vitali; writing: review and editing by Carmela Protano and Matteo Vitali; supervision by Matteo Vitali. All authors read and approved the final manuscript
Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request