Environ. Sci. Technol. 2006, 40, 7854-7860
Struggle for Quality in Determination of Perfluorinated Contaminants in Environmental and Human Samples S T E F A N P . J . V A N L E E U W E N , * ,† ANNA KA ¨ RRMAN,‡ BERT VAN BAVEL,‡ JACOB DE BOER,† AND GUNILLA LINDSTRO ¨ M‡ Institute for Environmental Studies (IVM), Free University Amsterdam, De Boelelaan 1085, 1081 HV, Amsterdam, The Netherlands, and Man-Technology-Environment Research Centre (MTM), O ¨ rebro University, SE-701 82 O ¨ rebro, Sweden
The first worldwide interlaboratory study on the analyses of 13 perfluorinated compounds (PFCs) in three environmental and two human samples indicates a varying degree of accuracy in relation to the matrix or analyte determined. The ability of 38 participating laboratories from 13 countries to determine the analytes in the various matrices was evaluated by calculation of z-scores according to the Cofino model. The PFCs which were reported most frequently by the laboratories, and assessed with the most satisfactory agreement, were perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA). In general, the level of agreement between the participating laboratories decreased in the following order: PFC standard solution (76% satisfactory z-scores of 3 years experience). This may be the key reason why better results were obtained in the human samples dataset as compared to the environmental samples, although the different nature of the samples and the different analytical methodology should not be neglected. Some laboratories indicated that they just started with method development and used this opportunity to evaluate their progress. Their methods were not fully validated which may have added substantially to the high variance in the datasets. However, principal component analysis (PCA) of the environmental data did not reveal a relation between the level of experience and satisfactory z-scores in FLE, FT, or water samples. MS/MS versus MS. Most laboratories used MS/MS for analysis of PFOS, which enables the detection of daughterions, of which m/z 80 and 99 were the most commonly measured. IT-MS can be used in MS/MS mode for perfluorinated carboxylic acids whereas for perfluorinated sulfonates the technique can only be applied in single MS mode. Single quadrupole MS laboratories (labs 20, 21, 24, and 35) and those using ion-trap MS (labs 15, 27, and 36) lack the ability to detect a daughter ion and may, therefore, suffer from mass interferences. However, from Figure 3, it cannot be concluded that there is a systematic bias from laboratories using single quadrupole compared to MS/MS laboratories as they are distributed over the dataset just as the MS/MS laboratories are. This does not exclude the occurrence of a mass interference, but other error sources may have a stronger effect on the total analytical error. Several MS/MS laboratories show a considerable bias from the assigned value, which presumably is due to the matrix effect discussed above. Laboratory 22 applied TOF-MS with sufficient resolution to avoid mass interferences. Generally, care should be taken to avoid interferences when using less sophisticated techniques, e.g., single quadrupole MS. PFOA. The agreement between the laboratories was less good for more complex matrices (fish tissue, fish liver extract, and water, Table 2). Also, fewer reports were submitted for
FIGURE 4. Relative standard deviations (rsd) of all reported results for 7 PFCs in all tested matrices. Rsd values were only calculated for datasets with g5 submitted results. these matrices. As with PFOS, the majority of the laboratories achieved good results for the SS (Table 2) and should not have serious problems with their calibration. Concerning the FLE and the FT samples, assigned and median values are close to the spiked values, although it should be noted that the true value is not exactly known since small amounts of PFOA were expected to be present in the FLE and FT samples prior to the fortification. Compared to PFOS, more laboratories produced satisfactory PFOA results for the FT sample (Table 2). This may be associated with the use of a perfluorinated carboxylic acid-type of internal standard (13C2PFOA, 13C2-PFDA, PFDoA, or 7H-PFHpA) by most of the laboratories. However, laboratories using no internal standard showed a performance similar to those using internal standards. Some of these laboratories have applied other corrections, e.g. recovery correction for losses during sample extraction and clean-up. The agreement of the 19 submitted results for the water sample was also poor for PFOA (minmax range of 3.4-190 ng/L). The low sample pH of 2 presumably led to a partial protonation the anion PFOA. This may have effected the solubility in the sample as well as the extraction/cleanup of the sample. Further research is underway exploring the effect of storage conditions of the water sample. The agreement between laboratories for the HP and HB samples was good with a percentage of satisfactory z-scores similar or even better than for the SS (Table 2). The minmax range is less than a factor of 10, which is better than for the environmental matrices included in this study. This indicates that currently applied extraction and cleanup methods for human blood are satisfactory as opposed to the methods used for environmental samples. Other PFC Compounds. A considerable number of laboratories submitted data on other PFCs such as PFOSA, PFHxS, PFHxA, PFNA, PFDA, and PFDoA. For all PFCs with >5 reported results, basic statistical analyses were performed. Limited data was provided for PFBA and PFDS. For PFDS no fully characterized standard was commercially available at the time this study was conducted, which hampered the quantification of this compound. Figure 4 shows the rsd values based on all submitted results (no outliers have been removed) for seven PFCs in all matrices included in this study. Regarding the SS results of PFHxS and PFHxA, there is reasonable agreement between the individual laboratories. For PFDoA the situation is not as good, probably due to the low concentration of 3.8 ng/mL in the SS. The high rsd value of PFOSA may be caused by the difficult deprotonation of this neutral compound in the LC-MS electrospray, and therefore, GC-MS is suggested as an alternative method (6). Concerning the FLE and the FT samples, the rsd values
increase as the matrix becomes more complex. This was also observed for PFOS and PFOA and is confirmed by the PFHxS, PFHxA, and PFNA results (Figure 4). The results for the water sample are poor; besides the reasons already discussed in the PFOS and PFOA section, the high rsd values may also result from the very low concentration levels in the sample. The results for human samples are generally in better agreement compared to the environmental samples. The rsd values of the HP and HB data range from 29 to 64% (excluding PFOSA), and are often close or even below the SS rsd values. The levels of PFHxA and were below the limit of quantification (LOQ) in all samples. The level of PFOSA was below LOQ for the HP sample. Summarizing, this study showed that the pool of participating laboratories were not able to produce consistent data, although individual laboratories may have long experience in the field and applied their very well validated methods. In some cases, interlaboratory results can highlight a relation between use of unoptimised techniques or methods and poor results. However, in many cases, laboratories may suffer from multiple difficulties, which hinder clear identification of the error sources. This has also been observed in interlaboratory studies on other emerging contaminants such as BFRs (11) The laboratories are recommended to critically assess their analytical procedures aiming at reducing possible sources of error. Issues to be assessed are (poor) extraction efficiency, suitability of external (or solvent) calibration, suitability of native PFCs as internal standards, quality of (internal) standards used, matrix effects (and need for cleanup steps to remove those), and selectivity of MS(/MS) technique. These issues have been discussed here and more information on quality issues in PFC analysis can be found elsewhere (6).
Acknowledgments We thank the participants for their contribution to this study. MD Olle Berse´us at the University Hospital of O ¨ rebro is acknowledged for supplying the blood and plasma sample material. Prof. Dr. Wim Cofino is thanked for his support in the statistical evaluation of the data. Mr. Marco Lohman, Evert van Barneveld, Christiaan Kwadijk, and Ike van der Veen are thanked for their skilful contributions to this study. The European Commission is thanked for the financial support of the Perforce project (www.science.uva.nl/perforce, NEST contract 508967). Plastics Europe is acknowledged for their financial support. QUANAS are acknowledged for supporting Mrs. Astrid Zammit’s stay at RIVO.
Supporting Information Available The list of participants, sample information, method information, participant results (including basic statistics), data VOL. 40, NO. 24, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7859
distribution plots and z-score plots are available. This material is available free of charge via the Internet at http:// pubs.acs.org.
Literature Cited (1) Hoff, P.; van de Vijer, K. I.; van Dongen, W.; Esmans, E. L.; Blust, R.; de Coen, W. Perfluorooctane sulfonic acid in Bib (trisopterus luscus) and Plaice (pleuronectes platessa) from the Belgian North Sea: Distribution and biochemical effects. Environ. Toxicol. Chem. 2003, 22, 608-614. (2) Kannan, K.; Tao, L.; Sinclair, E.; Pastva, S. D.; Jude, D. J.; Giesy, J. P. Perfluorinated compounds in aquatic organisms at various trophic levels in a great lakes food chain. Arch. Environ. Contam. Toxicol. 2005, 48, 559-566. (3) Martin, J. W.; Smithwick, M. M.; Braune, B. M.; Hoekstra, P. F.; Muir, D. C.; Mabury, S. A. Identification of long-chain perfluorinated acids in biota from the Canadian Arctic. Environ. Sci. Technol. 2004, 38, 373-380. (4) Yamashita, N.; Kannan, K.; Taniyasu, S.; Horii, Y.; Petrick, G.; Gamo, T. A global survey of perfluorinated acids in oceans. Mar. Pollut. Bull. 2005, 51, 658-668. (5) Kannan, K.; Corsolini, S.; Falandysz, J.; Fillman, G.; Kumar, K. S.; Loganathan, B. G.; Mohd, M. A.; Olivero, J.; van Wouwe, N.; Yang, J. H.; Aldous, K. M. Perfluorooctanesulfonate and related fluorochemicals in human blood from several countries. Environ. Sci. Technol. 2004, 38, 4489-4495. (6) Martin, J. W.; Kannan, K.; Berger, U.; de Voogt, P.; Field, J.; Franklin, J.; Giesy, J. P.; Harner, T.; Muir, D. C.; Scott, B.; Kaiser, M.; Ja¨rnberg, U.; Jones, K. C.; Mabury, S. A.; Schroeder, H.; Simcik, M.; Sottani, C.; van Bavel, B.; Ka¨rrman, A.; Lindstro¨m, G.; van Leeuwen, S. Analytical challenges hamper perfluoroalkyl research. Environ. Sci. Technol. 2004, 248A-255A. (7) de Boer, J.; Cofino, W. P. First world-wide interlaboratory study on polybrominated diphenylethers (PBDEs). Chemosphere 2002, 46. (8) Quevauviller, P.; Maier, E. A. In Interlaboratory studies and certified reference materials for environmental analysis, 1st ed.; Elsevier: Amsterdam, 1999; Vol. 22.
7860
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 24, 2006
(9) Cofino, W. P.; van Stokkum, I. H. M.; Wells, D. E.; Ariese, F.; Wegener, J. W. M.; Peerboom, R. A. L. A new model for the inference of population characteristics from experimental data using uncertainties. Application to interlaboratory studies. Chemom. Intell. Lab. Syst. 2000, 53, 37-55. (10) Wells, D. E.; Cofino, W. P. The assessment of the QUASIMEME laboratory performance studies data: Techniques and approach. Mar. Pollut. Bull. 1997, 35, 18-27. (11) de Boer, J.; Wells, D. E. Pitfalls in the analysis of brominated flame retardants in environmental, human and food samplesincluding results of three international interlaboratory studies. Trac-Trends Anal. Chem. 2006, 25, 364-372. (12) Hansen, K. J.; Clemen, L. A.; Ellefson, M. E.; Johnson, H. O. Compound-Specific, Quantitative Characterization of Organic Fluorochemicals in Biological Matrices. Environ. Sci. Technol. 2001, 35, 766-770. (13) Berger, U.; Haukas, M. Validation of a screening method based on liquid chromatography coupled to high-resolution mass spectrometry for analysis of perfluoroalkylated substances in biota. J. Chromatogr. A 2005, 210-217. (14) Karrman, A.; van Bavel, B.; Jarnberg, U.; Hardell, L.; Lindstrom, G. Development of a solid-phase extraction-HPLC/single quadrupole MS method for quantification of perfluorochemicals in whole blood. Anal. Chem. 2005, 77, 864-870. (15) Verreault, J.; Houde, M.; Gabrielsen, G. W.; Berger, U.; Haukas, M.; Letcher, R. J.; Muir, D. C. G. Perfluorinated alkyl substances in plasma, liver, brain and eggs from Glaucous Gulls (Larus hyperboreus) from the Norweigan Arctic. Environ. Sci. Technol. 2005, 39, 7439-7445. (16) Yamashita, N.; Kannan, K.; Taniyasu, S.; Horii, Y.; Okazawa, T.; Petrick, G.; Gamo, T. Analysis of perfluorinated acids at partsper-quadrillion levels in seawater using liquid chromatographytandem mass spectrometry. Environ. Sci. Technol. 2004, 38, 5522-5528.
Received for review May 3, 2006. Revised manuscript received July 29, 2006. Accepted September 24, 2006. ES061052C
