Stefanie
Nübler
a,
Therese
Burkhardt
b,
Moritz
Schäfer
a,
Johannes
Müller
a,
Karin
Haji-Abbas-Zarrabi
a,
Nikola
Pluym
b,
Max
Scherer
b,
Gerhard
Scherer
b,
Marta
Esteban-López
c,
Argelia
Castaño
c,
Hans G. J.
Mol
d,
Holger M.
Koch
e,
Jean-Philippe
Antignac
f,
Jana
Hajslova
g,
Katrin
Vorkamp
h and
Thomas
Göen
*a
aFriedrich-Alexander-Universität Erlangen-Nürnberg, Institute and Outpatient Clinic of Occupational, Social, and Environmental Medicine, Henkestraße 9-11, 91054 Erlangen, Germany. E-mail: thomas.goeen@fau.de
bABF Analytisch-Biologisches Labor, ABF GmbH, Semmelweisstr. 5, 82152 Planegg, Germany
cNational Center for Environmental Health, Instituto de Salud Carlos III, Ctra. Majadahonda – Pozuelo km 2.2, 28220 Madrid, Spain
dWageningen Food Safety Research, Wageningen University and Research, Akkermaalsbos 2, Wageningen, 6708 WB, Netherlands
eInstitute for Prevention and Occupational Medicine of the German Social Accident Insurance, Institute of the Ruhr-University Bochum (IPA), Bürkle-de-la-Camp-Platz 1, 44789 Bochum, Germany
fOniris, INRAE, UMR 1329 Laboratoire d'Etude des Résidus et Contaminants dans les Aliments (LABERCA), F-44307 Nantes, France
gDepartment of Food Analysis and Nutrition, University of Chemistry and Technology Prague, Technicka 5, 16000 Prague, Czech Republic
hDepartment of Environmental Science, Aarhus University, Frederiksborgvej 399, 4000 Roskilde, Denmark
First published on 12th November 2024
Exposure to aromatic amines may occur via tobacco smoke, hair dyes or tattoo inks, but also in the workplace during certain manufacturing processes. As some aromatic amines are known or suspected carcinogens, human biomonitoring (HBM) is essential to assess their exposure. Aromatic amines were among the selected chemicals in HBM4EU, a European-wide project to harmonise and advance HBM within 30 European countries. For this purpose, the analytical comparability and accuracy of participating laboratories were assessed by a QA/QC programme comprising interlaboratory comparison investigations (ICIs) and external quality assurance schemes (EQUASs). This paper presents the evaluation process and discusses the results of three ICI/EQUAS rounds for the determination of aromatic amines in urine conducted in 2019 and 2020. The final evaluation included ten participants which analysed the following six targeted aromatic amines over three rounds: aniline, ortho-toluidine (TOL), 4,4′-methylenedianiline (MDA), 4,4′-methylenebis(2-chloroaniline) (MOCA), 2,4-diaminotoluene (2,4-TDA), and 2,6-diaminotoluene (2,6-TDA). Most participants achieved satisfactory and highly comparable results, although low quantification limits were required to quantify the parameters at the level of exposure in the general population. Hydrolysis of the sample followed by liquid–liquid extraction and subsequent analysis of the derivatised analytes by means of GC-MS/MS were preferred for the sensitive and precise determination of aromatic amines in urine. This QA/QC programme succeeded in establishing a network of laboratories with high analytical comparability and accuracy for the analysis of aromatic amines in Europe.
High acute exposure to aromatic amines can result in the production of methaemoglobin and thus may cause methaemoglobinemia.8 The most significant health concern is, however, the development of cancer caused by chronic exposure to aromatic amines.9
Due to the different exposure routes, human biomonitoring (HBM) is a powerful tool for the holistic assessment of chemical exposure and risk.10 In case of exposure to aromatic amines and their precursors, the determination of aromatic amines (free or conjugated) in urine is a well-established HBM approach.11
Aromatic amines were among the chemicals selected as first-priority substances in the European HBM Initiative HBM4EU.12 This European-wide project was a joint effort of 30 countries and European Commission authorities, with the aims to harmonise and advance HBM in Europe and to improve risk management for chemicals in support of policymaking.13–15 A major objective of the HBM4EU project was to establish a network of analytical laboratories across Europe that would produce high-quality and comparable HBM data for the prioritised substances.16 A comprehensive Quality Assurance/Quality Control (QA/QC) scheme was designed consisting of several rounds of interlaboratory comparison investigations (ICIs) and external quality assurance schemes (EQUASs) for all prioritised substances. ICIs investigated the comparability of results between laboratories participating in the HBM4EU QA/QC programme. EQUASs were based on the comparison of the participants' results with those of selected expert laboratories that applied validated analytical methods and had experience with HBM of aromatic amines in population studies.16
This paper presents the QA/QC programme developed in HBM4EU for a range of aromatic amines, including the evaluation process, the results obtained, and the main challenges encountered. Originally, nine biomarkers of aromatic amines were included in the programme. However, due to the low participation in the first round, three biomarkers (p-aminophenol, N-acetyl-4-aminophenol, and p-phenylenediamine) were excluded. The final biomarkers were: aniline, ortho-toluidine (TOL), 4,4′-methylenedianiline (MDA), 4,4′-methylenebis(2-chloroaniline) (MOCA), 2,4-diaminotoluene (2,4-TDA), and 2,6-diaminotoluene (2,6-TDA) (Table 1). The structures of these six aromatic amines analysed in the ICI/EQUAS are shown in ESI Fig 1.† These compounds were identified as suitable biomarkers in HBM4EU, including substances with substantial (MDA, MOCA) and insufficient HBM information.12 Aniline and TOL are widely present in the environment and general biomarkers of exposure to aromatic amines, while MDA, MOCA, 2,4-TDA and 2,6-TDA are mainly associated with exposure to plastic products.11,12
Abbreviation | Target biomarker | CAS no. | Formula |
---|---|---|---|
a Other name: 2-methylaniline (2-MA). | |||
TOL | ortho-Toluidinea | 95-53-4 | C7H9N |
Aniline | Aniline | 62-53-3 | C6H7N |
MOCA | 4,4′-Methylenebis(2-chloroaniline) | 101-14-4 | C13H12Cl2N2 |
2,4-TDA | 2,4-Diaminotoluene | 95-80-7 | C7H10N2 |
2,6-TDA | 2,6-Diaminotoluene | 823-40-5 | C7H10N2 |
MDA | 4,4′-methylenedianiline | 101-77-9 | CH2(C6H4NH2)2 |
In response to two calls for laboratories to perform analysis of aromatic amines in HBM4EU, 18 laboratories expressed their interest, of which eleven laboratories (61%) from four countries (ESI Table 1†) finally registered for participation in the ICI/EQUAS programme. Participation was possible for the whole set of aromatic amine biomarkers (six) or for less. Laboratories were asked to report the measured concentrations of the CMs alongside the respective limits of quantification (LOQs).
For EQUAS evaluation, five expert laboratories had been selected prior to the ICI/EQUAS rounds according to selection criteria predetermined by the HBM4EU Quality Assurance Unit (QAU). The selection of these experts was based on the fact that these laboratories had experience in the determination of aromatic amines which was documented in peer-reviewed publications. In addition, the following selection criteria were considered: number of years of experience in the analysis of aromatic amines in urine, application of highly sensitive and selective analytical techniques with sufficiently low LOQs, application of isotope-labelled standards for quantification, availability of in-house validation reports, data on on-going intra-laboratory performance, ISO 17025 accreditation and success rate in round robin tests or comparative results in HBM studies. Four of the selected expert laboratories were from Germany and one was from the UK. All expert laboratories were also included as participants in the respective ICI/EQUAS rounds.
Two different pools of non-smoker urine were mixed, adjusted to pH 4.0, and spiked with a stock solution to obtain the analytes in the final concentrations shown in ESI Table 2.† The stock solution was prepared from individual standards of different suppliers. As the stability of the free aromatic amines, except for TOL, is poor, aniline, MOCA, 2,4-TDA, 2,6-TDA, and MDA were spiked as diacetyl-conjugates (synthesised and provided by Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany) and the participants were instructed to include a hydrolysis step that is typically carried out in the measurement of real samples.19 For spiking of aniline, free aniline (no. 51788-1ML-F, Merck KGaA, Darmstadt, Germany) was applied in the first round while acetanilide (no. 397237, Merck KGaA, Darmstadt, Germany) was used in the second and the third round. TOL was considered stable at pH 4.0 and was therefore not spiked as a conjugate (no. 397237, Merck KGaA, Darmstadt, Germany). Each CM was aliquoted to a volume of 8 mL in 15 mL falcon tubes (polypropylene, Greiner AG, Kremsmünster, Austria) and stored at ≤−20 °C until transport.
For homogeneity testing, ten tubes of CMlow and CMhigh, respectively, were randomly selected from the freezer (−20 °C) shortly after CM preparation. Samples were thawed, homogenised by vortex shaking, and analysed in duplicate. Results obtained were evaluated according to Fearn and Thompson20 as well as Thompson et al.21 A relative standard deviation (RSD) of 25% was chosen to be acceptable for homogeneity for CMlow and CMhigh, respectively. This threshold value was regarded as fit-for-purpose taking into account what is technically feasible and realistic in current routine practises with respect to the analytical method and concentration levels analysed.
For stability testing, three samples of each CM level were randomly selected from the freezer (−20 °C) and analyzed at t = 0 days (day of CM preparation) and on the day on which results were submitted (t = 40 days, 46 days or 60 days). Stability was assessed by comparing the means of the triplicates. Results were evaluated according to ISO 13528 (ref. 22) and the International Harmonised Protocol for the Proficiency Testing of Analytical Laboratories.21
A schematic overview of the requirements, assessment values and evaluation in an ICI or EQUAS is shown in Fig. 1. For an ICI, the robust mean of participants' results was taken as a consensus value (C) if at least seven quantitative results from participating laboratories were available and the uncertainty of the consensus was within certain requirements (details see ref. 16). Robust statistics (details see ref. 24 and 25) were applied to reduce the influence of outliers on C. For an EQUAS, a minimum of three designated expert laboratories analysed six samples of each CM; their means were used to calculate the assigned value (A) as a mean-of-means when the uncertainty of A (u), calculated as relative standard deviation (RSDexperts) divided by the square root of the number of expert laboratories, did not exceed 17.5%. Due to the low number of quantitative results, classical statistics were used in EQUASs. EQUAS evaluations had the advantage that participants' performance could also be assessed when the number of participants was too small or their results were too heterogeneous for ICI evaluation.
Z-scores of the participants' results (x) were calculated as a measure of proficiency using C or A and the target RSD (σT) of 25% (see equation in Fig. 1). The value for the highest variability (σT = 25%) considered acceptable for participants' results was set according to expert judgement, taking into account what is technically feasible in current routine practice.16
Only the absolute values of Z-scores were relevant for the assessment of laboratory performance and were categorised as follows:
• |Z| ≤ 2 ⇒ satisfactory result
• 2 < |Z| < 3 ⇒ questionable result
• |Z| ≥ 3 ⇒ unsatisfactory result.
For each participant, a round was considered “passed” for the specific biomarker if only satisfactory Z-scores were obtained in both CMs.
Aromatic amine | CM | Maximum number of participants (of which experts) | Comparability of results (study RSDR in %) | Z-score (%) | ||
---|---|---|---|---|---|---|
Satisfactory | Questionable | Unsatisfactory | ||||
a CM = control material; na = not applicable; study RSDR = robust relative standard deviation of participants' results. | ||||||
TOL | Low | 6 (3) | 17.7 | 94.4% | 0% | 5.6% |
High | 6 (3) | 12.9 | 94.4% | 5.6% | 0% | |
Aniline | Low | 5 (3) | na | na | na | na |
High | 5 (3) | 22.5 | 100% | 0% | 0% | |
MOCA | Low | 9 (5) | 52.4 | 83.8% | 8.3% | 7.8% |
High | 9 (5) | 34.4 | 83.8% | 8.3% | 7.8% | |
2,4-TDA | Low | 8 (5) | 22.1 | 86.5% | 4.7% | 8.8% |
High | 8 (5) | 20.4 | 86.5% | 4.7% | 8.8% | |
2,6-TDA | Low | 8 (5) | 61.4 | 77.8% | 4.2% | 18.0% |
High | 8 (5) | 45.1 | 82.3% | 4.7% | 13.0% | |
MDA | Low | 10 (5) | 16.7 | 96.3% | 0% | 3.7% |
High | 10 (5) | 16.6 | 92.2% | 4.2% | 3.7% |
The participants' results are summarised in Table 2 and presented in detail in ESI Table 7,† the latter containing the evaluation schemes, the evaluation results and the percentage of satisfactory, questionable, and unsatisfactory Z-scores in each round. Overall, the participants' results in each of the three ICI/EQUAS rounds were predominantly satisfactory. The percentage average of the number of satisfactory results (with |Z| ≤ 2, see Section 2.4) over all evaluable aromatic amines ranged from 62.5% to 100%, and increased slightly from round 1 to round 3 for CMlow and CMhigh (ESI Fig. 3†). Although the average percentage of satisfactory results for all parameters was almost the same across all rounds (88%), the increase in satisfactory results was higher for CMs with higher levels of aromatic amines (14.5% more satisfactory results in round 3 than in round 1) than for CMs with lower levels (8.4% increase). The training effect was therefore more pronounced at the higher concentrations, although the absolute values of the Z-scores did not significantly differ between CMlow and CMhigh (Mann–Whitney U test, U = 6102, Z = −1.112, p = 0.267).
Over all rounds, the participants achieved the most comparable results for TOL and MDA, as the average of the robust relative standard deviation of the participants' results (study RSDR) was below 20% for both aromatic amines (TOL: 15.3%, MDA: 16.6%). For 2,4-TDA and aniline (only one round), the mean study RSDR over all rounds was <23%, while it was clearly higher for MOCA (43.4%) and 2,6-TDA (53.3%). The determination of the lower levels of aromatic amines resulted in a higher mean study RSDR (34.0%) compared to the participants' results for the higher levels (mean study RSDR = 25.4%), which was to be expected given higher uncertainties close to the LOQs. Interestingly, the results of the participants varied the most in the 3rd round (mean study RSDR = 37.6%), while the mean study RSDR was lower in the 1st round (31.3%) and lowest in the 2nd round (19.5%). The third round might have been the most challenging one for the following two reasons. First, both CMs contained on average lower final concentrations of aromatic amines in the third round (mean = 30.2 ng mL−1) than in the first (58.3 ng mL−1) and the second round (62.6 ng mL−1). Second, only one sample of each CM was sent to the participants in the last round, so that repetition variances were not balanced, as was possible in the 1st and the 2nd round. Nevertheless, this challenging last round was passed with the best Z-scores of all rounds by most participants, which could reflect a certain training effect. The high variation of results in the third round was mainly caused by strongly deviating results from different individual participants for the biomarkers TOL, MOCA and 2,6-TDA. In contrast, the variation of the participants' results in the first round was not influenced by strongly deviating individual results. In the second round, only one participant reported very different results for MDA. However, by applying robust statistics, as described in Section 2.4, the influence of individual outliers on the evaluation could be reduced.
The relative standard deviation of the expert results (RSDexperts, 11.7%) was on average less than half the mean study RSDR (29.7%) across all rounds and CMs, which illustrates the expertise of the expert laboratories in measuring the respective aromatic amines, but might also be influenced by the fact that the experts could report mean values of six samples. However, not for all measurements, the uncertainty of the experts was low enough to enable an EQUAS evaluation (see ESI Table 7†).
The LOQs reported by the participants (including experts) ranged from 0.010 to 25.0 ng mL−1, being lowest for TOL and highest for 2,4- and 2,6-TDA, with a substantial range for 2,4- and 2,6-TDA (ESI Table 8†). Over the three rounds, most laboratories reported constant LOQs, two participants reported higher LOQs in successive rounds (for aniline or for MOCA, 2,4-TDA, 2,6-TDA, and MDA), while three participants could improve their LOQ for one biomarker each (TOL, MOCA, or MDA). The recommended LOQs according to the literature (0.05 ng mL−1 for TOL, 0.5 ng mL−1 for aniline, and 1 ng mL−1 for 2,4-TDA, 2,6-TDA, MDA and MOCA) were not achieved by all participants. For MDA, all participants met the requirements, and for 2,4-TDA and 2,6-TDA, 83% of the measurements achieved the required LOQs. In contrast, only around 50% of the reported LOQs for MOCA, aniline, and TOL were in line with the given requirements.
Significant differences were found in the absolute values of the Z-scores for different methods applied. The Z-scores of the HPLC measurements were significantly higher (p < 0.01) than those of the GC measurements (Fig. 2a). This is consistent with the finding that participants using GC to detect aromatic amines also reported significantly lower (p < 0.01) LOQs than participants using HPLC (Fig. 2b). In the GC measurements, no statistical difference was observed between the Z-scores obtained with single-quadrupole and triple-quadrupole detection (Mann–Whitney U-test, U = 2872, Z = −1.361, p = 0.175) or between the Z-scores obtained with PFPA or HBFA as a derivatisation reagent (Mann–Whitney U-test, U = 3036, Z = −1.426, p = 0.154). Overall, GC proved to be more sensitive and reliable than HPLC for detecting the six aromatic amines investigated in the ICI/EQUAS.
In order to compare the two different evaluation schemes, the consensus values of the participants' results were also calculated for cases in which an assessment as EQUAS was possible. The relative difference of the mean values obtained by ICI (C) and EQUAS (A) calculated as [(C − A)/A] × 100% was predominantly ≤10% and, interestingly, it was greater in the high-level CMs than in the lower ones (Table 2). On average, the relative difference between A and C was only 7.5%, which shows that ICI and EQUAS generally lead to comparable Z-scores and enabled an equivalent evaluation of participants' performance for the measurement of the respective aromatic amines.
A | Assigned value |
C | Consensus value |
CM | Control material |
EQUAS | External quality assurance scheme |
HFBA | Heptafluorobutyric anhydride |
HBM | Human biomonitoring |
HBM4EU | European human biomonitoring initiative |
ICI | Interlaboratory comparison investigation |
ISTD | Internal standard |
LOQ | Limit of quantification |
MDA | 4,4′-Methylenedianiline |
MOCA | 4,4′-Methylenebis(2-chloroaniline) |
TOL | ortho-Toluidine |
PFPA | Pentafluoropropionic anhydride |
RSD | Relative standard deviation |
study RSDR | Robust relative standard deviation of participants' results |
2,4-TDA | 2,4-Diaminotoluene |
2,6-TDA | 2,6-Diaminotoluene |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ay01309c |
This journal is © The Royal Society of Chemistry 2025 |