INTRODUCTION - terms of
reference
In the follow-up to its Recommendation of 1st of July 1998 on phthalates in toys and childcare articles, the Commission is, inter alia, monitoring developments in the field of phthalate migration test methods.
The CSTEE, in its opinion of 27th November 1998 on "Phthalate migration from soft PVC toys and childcare articles", has recommended that "an inter-laboratory comparison exercise be carried out in order to document the reproducibility of the Dutch laboratory method to measure phthalate release from PVC toys".
A report has been recently published by TNO on "Validation of the method: Determination of Diisononylphthalate in saliva simulant" - TNO report v99.598 of 27th May 1999 (CSTEE/97/1-Add 149).
Shortly after, in June 1999, the LGC has published a report (LGC Technical Report n° LGC/1999/DTI/004 on Inter-laboratory Validation of Laboratory-Based Agitation Methods for Determination of Phthalate Plasticiser migration from toys and childcare articles (CSTEE/97/1-Add 148D).
In relation to the above-mentioned reports, the opinion of the CSTEE has been requested on the questions presented below (The answers are in Italics in direct connection to the individual questions).
In addition, the CSTEE has been requested to consider additional results of in-vivo tests in the US and the relation between in-vivo and in-vitro tests (CSTEE/97/1-Add 163A-D).
TNO report
1. Has the inter-laboratory comparison in question been performed using a target value for DINP release of 9 µg/10cm2/min, as it was requested by the CSTEE?
No, the method developed by TNO is aiming at a migration of about 1.38 µg/10 cm2/min, which was the mean release found in the Dutch in-vivo study (Könemann, W.H., ed., RIVM report 613320 002, CSTEE/97/1-Add 107).
2. How do the results of the tests performed with the method in question compare with the "in vivo" data available from the Dutch and other studies?
The overall mean DINP migration found by the five laboratories for the "standard PVC discs" is somewhat higher (2.09 µg/10cm2/min) than the mean value found in the in-vivo study (1.38 µg/ 10 cm2/min).
The overall mean DINP migration found by the five laboratories for "discs from an unidentified teething ring" is somewhat higher (2.40 µg/10 cm2/min) than the mean value found in the in-vivo study (1.63 µg/10 cm2/min).
3. The variability of mean values for release of DINP (table 13 of the report) obtained for a same product in the various laboratories goes from a minimum factor 1.9 (sample 4) to a maximum factor 4.2 (sample 9). What may be the reasons for such variability and what are the implications in terms of possible "false positive" or "false negative" when the migration level is near the limit set for control purposes?
The large differences in results, both for the standard PVC discs and for the discs from the test samples, reflects a poor reproducibility among the laboratories. This will be a serious problem if the method is used for legal purposes.
4. Do the results of a validation study of the TNO method satisfy the CSTEE that this method could be used in routine enforcement to assure that the guidance value for DINP set down in their Opinion can be respected?
The answers given under questions 1 and 3 explain why the CSTEE does not accept this method for testing of plasticiser migration from toys.
LGC report
1. How do the "migration trial results" for commercial samples obtained in the series of tests performed at 37°C and 65°C compare with the target value of 9 µg/10cm2/min and the in vivo data available from the Dutch and other studies?
The LGC experiments are performed using "reference discs" different from those used in the Dutch in-vivo study. It is mentioned in the text that "the LGC reference discs have already been shown to demonstrate similar release characteristics to the Dutch standard discs", but this documentation is not available to the CSTEE. Bearing this in mind, it can be said that the methods LGC have set up give mean values of DINP migration from "LGC standard PVC discs" 1.43 µg/10 cm2/min at 37°C, which is close to the mean value found in the Dutch in-vivo study; and 8.83 µg/10 cm2/min at 65°C, which is close to the highest migration found in the in- vivo study.
2. Are the samples used representative of toys and childcare articles intended to be mouthed by children under 3 years?
The CSTEE cannot answer this question because the report does not contain that information.
3. The variability of the mean values for release of DINP at 65°C (Table 6 of the report) obtained for a same product in the various laboratories goes from a minimum factor of 1.8 (sample 4) to a maximum factor of 2.8 (sample 3), whereas some results are apparently missing. The variability seems to change unpredictably with the temperature [e.g. for sample 4: Lab.1 Õ 2.74 (37°C), 2.86 (65°C); Lab.3 Õ 1.16 (37°C), 5.3 (65°C)]. What may be the reasons for this variability and what are the consequences in terms of possible "false positive" or "false negative" when migration level is near the limit set for control purposes? How should the fact that certain results are missing affect the conclusions?
The CSTEE has also observed the missing data in those and other tables but has no explanation for these omissions. The variability both within and between laboratories is large making it difficult to use the method for legal purposes. The present analysis is obviously a difficult determination and variability may be due to a situation where migration gives phthalate concentrations in the aqueous phase which are higher than the solubility of DINP so that micelles are probably formed.
4. Do the results of the validation study of the LGC "simulated" method satisfy the CSTEE that this method could be used in routine enforcement to assure that the guidance value for DINP set down in their Opinion can be respected?
This method is developed to give migrations close to the mean value found in the Dutch in-vivo study and cannot be used to simulate the maximum DINP migration (ca. 9 µg/10 cm2/min) found in that study and which the CSTEE requested in its Opinion.
5. Do the results of the validation study of the LGC "stringent" method satisfy the CSTEE that this method could be used in routine enforcement to assure that the guidance value for DINP set down in their Opinion can be respected?
This method is developed to mimic the worst case scenario in the in-vivo study and the mean value of several tests on standard PVC disc in five different laboratories comes close to that. However, the variations in the results both within and among the laboratories are so high that it is not possible to use the method for regulatory purposes in its present status.
Very few data are available on discs from test samples. It appears that the participating laboratories have analysed only a single disc from each sample. So, the repeatability (variation in test-results within the laboratories) data on the samples are not available. The reproducibility (variation in test-results among the laboratories) is also very high, similar to that for standard PVC-discs. Therefore, the method, as reported, is not suitable for the purpose of enforcement.
Both in-vitro methods
1. These methods seem conceived to achieve predetermined extraction (or migration) levels. Is it possible to calibrate and use these methods in order to discriminate among commercial products giving rise "in vivo" to a migration below/above the recommended limit value?
a) The TNO method is validated for average migration of DINP (approximately 2 m g/min) found in the in-vivo study performed by Dutch Consensus Group. By this method it may not be possible to identify commercial products which may release ³ 6.7 m g/min DINP corresponding to the TDI of DINP.
b) The LGC 'stringent method' did reveal in-vitro migration of > 6.7 m g/min DINP from 'LGC standard PVC disc'. However, there is no documentation available to confirm that the 'LGC standard PVC disc' is equivalent to the standard PVC disc used in the Dutch Consensus Study. Finally, the data produced for the validation of the test method is very limited and the repeatability and the reproducibility of the test method are poor. Therefore, the method in the present state may not be able to discriminate among commercial products giving rise "in- vivo" to a migration below/above the recommended limit value.
2. Should these methods be subjected to a formal ISO ring test before they can be considered as validated? Can otherwise the methods be considered as validated and if not, which additional efforts would be necessary? The committee is invited to make specific recommendations, where relevant, in order to improve the methods in view of achieving this.
No, it would be a waste of resources to set up an ISO ring test at the present state.
From the studies reported so far, it appears that in-vitro migration rate of phthalates from a PVC product depends upon two factors, the matrix and the conditions employed for the forced migration. The matrix related factors - phthalate content, homogeneity of the phthalate distribution in the matrix, other matrix constituents - all seem to influence migration rate. The test conditions which have been shown to influence in-vitro forced migration of phthalates from PVC-products appear to be: mechanical force applied, temperature and the saliva composition. Besides the above mentioned factors, the physical properties of a PVC-product, as a result of 'moulding' and 'curing' processes used for the manufacture of PVC products, may also affect the migration rate of the plasticiser.
It is suggested that for the validation of a test method for phthalate migration from commercial products:
- The target migration rate for the test method should be 9m g/min from the reference discs used in the Dutch volunteer study. This will allow to identify the products which may not comply with the recommended limit values. The test method, prior to a validation exercise using ISO guidelines, should demonstrate that products with DINP migration rates below/above the recommended limit value can be discriminated.
- The test samples should be products of a known matrix with respect to manufacturing process and other matrix factors described above. Test samples should include products produced using different manufacturing process and with different phthalate contents. The reference material should be available upon request by national authorities.
- The simulation conditions used for the test method for in-vitro migration should in principle be as close as possible to in-vivo conditions. It is possible to change the extracted amount by changing at least three different parameters: the physical force, the temperature and the composition of the saliva simulant. In the LGC methods the two first were used to "calibrate" against the target values, but the repeatability within a laboratory and the reproducibility among laboratories were poor. It may be worth to try to use different saliva simulant composition to see whether that gives less variations. To match the in-vivo studies, an artificial saliva with some organic content should be preferred.
- The metal balls used as a source of mechanical force in the LGC study may tear off the PVC sample. This process will result in poor repeatability. In the LGC report, no mention has been made concerning the physical appearance of the test samples after the completion of the experiment. Moreover, the particles release due to aberration of PVC may also affect the test results. In the report from JRC Ispra (CSTEE/ 97/1 - Add. 170), a reference has been made to a home made 'gnawing machine', which may be a better option as a source of mechanical force for the in-vitro test method.
- The phthalate analysis should be performed by an analytical technique which can identify phthalates unequivocally.
- The variability of the test method should be as low as possible, ideally < 20% (Allowance of 20% tolerance in the results permits 40% deflection from the target value!)
Final recommendation
- For further work with migration studies, the use of a standardised reference material is recommended. Such a material should be of the same composition as the one used in the Dutch in vivo experiment.
The US in-vivo and in-vitro studies
1. How does the CSTEE interpret the additional results of in-vivo tests in the US (CSTEE/97/1-Add.163/C) and the relation between in-vivo and in-vitro tests.
In the US in-vivo test ten persons chewed/mouthed discs cut from a toy containing almost 43% DINP, and the saliva was collected and analysed for phthalates. The procedure was divided in four times fifteen minutes and the saliva samples from these periods were analysed separately. The results show variations for each subject between the four periods (up to a factor of 3) and a variation between subjects of almost an order of magnitude. The average of all results was 4.3 µg/10 cm2/min and the maximum value was 9.7 µg/10 cm2/min, with an extreme for one 15-minute period of 13.4 µg/10 cm2/min. These results support those obtained in the Dutch and Austrian studies used for the earlier opinion of the CSTEE.
The in-vitro migration of DINP was also tested in the US study. Different sizes of pistons were used to mechanically treat the test material in different saliva simulants. The material used in the in- vivo test gave a migration of 3.3 µg/10 cm2/min with extremes of 2.5 and 4.1, respectively, in a series of five determinations. The in-vitro method produced a mean close to the mean value from in-vivo study, but the repeatability seem to be rather low (few data points) and it is not possible to say anything about the reproducibility among different laboratories.
In the follow-up to its Recommendation of 1st of July 1998 on phthalates in toys and childcare articles, the Commission is, inter alia, monitoring developments in the field of phthalate migration test methods.
The CSTEE, in its opinion of 27th November 1998 on "Phthalate migration from soft PVC toys and childcare articles", has recommended that "an inter-laboratory comparison exercise be carried out in order to document the reproducibility of the Dutch laboratory method to measure phthalate release from PVC toys".
A report has been recently published by TNO on "Validation of the method: Determination of Diisononylphthalate in saliva simulant" - TNO report v99.598 of 27th May 1999 (CSTEE/97/1-Add 149).
Shortly after, in June 1999, the LGC has published a report (LGC Technical Report n° LGC/1999/DTI/004 on Inter-laboratory Validation of Laboratory-Based Agitation Methods for Determination of Phthalate Plasticiser migration from toys and childcare articles (CSTEE/97/1-Add 148D).
In relation to the above-mentioned reports, the opinion of the CSTEE has been requested on the questions presented below (The answers are in Italics in direct connection to the individual questions).
In addition, the CSTEE has been requested to consider additional results of in-vivo tests in the US and the relation between in-vivo and in-vitro tests (CSTEE/97/1-Add 163A-D).
TNO report
1. Has the inter-laboratory comparison in question been performed using a target value for DINP release of 9 µg/10cm2/min, as it was requested by the CSTEE?
No, the method developed by TNO is aiming at a migration of about 1.38 µg/10 cm2/min, which was the mean release found in the Dutch in-vivo study (Könemann, W.H., ed., RIVM report 613320 002, CSTEE/97/1-Add 107).
2. How do the results of the tests performed with the method in question compare with the "in vivo" data available from the Dutch and other studies?
The overall mean DINP migration found by the five laboratories for the "standard PVC discs" is somewhat higher (2.09 µg/10cm2/min) than the mean value found in the in-vivo study (1.38 µg/ 10 cm2/min).
The overall mean DINP migration found by the five laboratories for "discs from an unidentified teething ring" is somewhat higher (2.40 µg/10 cm2/min) than the mean value found in the in-vivo study (1.63 µg/10 cm2/min).
3. The variability of mean values for release of DINP (table 13 of the report) obtained for a same product in the various laboratories goes from a minimum factor 1.9 (sample 4) to a maximum factor 4.2 (sample 9). What may be the reasons for such variability and what are the implications in terms of possible "false positive" or "false negative" when the migration level is near the limit set for control purposes?
The large differences in results, both for the standard PVC discs and for the discs from the test samples, reflects a poor reproducibility among the laboratories. This will be a serious problem if the method is used for legal purposes.
4. Do the results of a validation study of the TNO method satisfy the CSTEE that this method could be used in routine enforcement to assure that the guidance value for DINP set down in their Opinion can be respected?
The answers given under questions 1 and 3 explain why the CSTEE does not accept this method for testing of plasticiser migration from toys.
LGC report
1. How do the "migration trial results" for commercial samples obtained in the series of tests performed at 37°C and 65°C compare with the target value of 9 µg/10cm2/min and the in vivo data available from the Dutch and other studies?
The LGC experiments are performed using "reference discs" different from those used in the Dutch in-vivo study. It is mentioned in the text that "the LGC reference discs have already been shown to demonstrate similar release characteristics to the Dutch standard discs", but this documentation is not available to the CSTEE. Bearing this in mind, it can be said that the methods LGC have set up give mean values of DINP migration from "LGC standard PVC discs" 1.43 µg/10 cm2/min at 37°C, which is close to the mean value found in the Dutch in-vivo study; and 8.83 µg/10 cm2/min at 65°C, which is close to the highest migration found in the in- vivo study.
2. Are the samples used representative of toys and childcare articles intended to be mouthed by children under 3 years?
The CSTEE cannot answer this question because the report does not contain that information.
3. The variability of the mean values for release of DINP at 65°C (Table 6 of the report) obtained for a same product in the various laboratories goes from a minimum factor of 1.8 (sample 4) to a maximum factor of 2.8 (sample 3), whereas some results are apparently missing. The variability seems to change unpredictably with the temperature [e.g. for sample 4: Lab.1 Õ 2.74 (37°C), 2.86 (65°C); Lab.3 Õ 1.16 (37°C), 5.3 (65°C)]. What may be the reasons for this variability and what are the consequences in terms of possible "false positive" or "false negative" when migration level is near the limit set for control purposes? How should the fact that certain results are missing affect the conclusions?
The CSTEE has also observed the missing data in those and other tables but has no explanation for these omissions. The variability both within and between laboratories is large making it difficult to use the method for legal purposes. The present analysis is obviously a difficult determination and variability may be due to a situation where migration gives phthalate concentrations in the aqueous phase which are higher than the solubility of DINP so that micelles are probably formed.
4. Do the results of the validation study of the LGC "simulated" method satisfy the CSTEE that this method could be used in routine enforcement to assure that the guidance value for DINP set down in their Opinion can be respected?
This method is developed to give migrations close to the mean value found in the Dutch in-vivo study and cannot be used to simulate the maximum DINP migration (ca. 9 µg/10 cm2/min) found in that study and which the CSTEE requested in its Opinion.
5. Do the results of the validation study of the LGC "stringent" method satisfy the CSTEE that this method could be used in routine enforcement to assure that the guidance value for DINP set down in their Opinion can be respected?
This method is developed to mimic the worst case scenario in the in-vivo study and the mean value of several tests on standard PVC disc in five different laboratories comes close to that. However, the variations in the results both within and among the laboratories are so high that it is not possible to use the method for regulatory purposes in its present status.
Very few data are available on discs from test samples. It appears that the participating laboratories have analysed only a single disc from each sample. So, the repeatability (variation in test-results within the laboratories) data on the samples are not available. The reproducibility (variation in test-results among the laboratories) is also very high, similar to that for standard PVC-discs. Therefore, the method, as reported, is not suitable for the purpose of enforcement.
Both in-vitro methods
1. These methods seem conceived to achieve predetermined extraction (or migration) levels. Is it possible to calibrate and use these methods in order to discriminate among commercial products giving rise "in vivo" to a migration below/above the recommended limit value?
a) The TNO method is validated for average migration of DINP (approximately 2 m g/min) found in the in-vivo study performed by Dutch Consensus Group. By this method it may not be possible to identify commercial products which may release ³ 6.7 m g/min DINP corresponding to the TDI of DINP.
b) The LGC 'stringent method' did reveal in-vitro migration of > 6.7 m g/min DINP from 'LGC standard PVC disc'. However, there is no documentation available to confirm that the 'LGC standard PVC disc' is equivalent to the standard PVC disc used in the Dutch Consensus Study. Finally, the data produced for the validation of the test method is very limited and the repeatability and the reproducibility of the test method are poor. Therefore, the method in the present state may not be able to discriminate among commercial products giving rise "in- vivo" to a migration below/above the recommended limit value.
2. Should these methods be subjected to a formal ISO ring test before they can be considered as validated? Can otherwise the methods be considered as validated and if not, which additional efforts would be necessary? The committee is invited to make specific recommendations, where relevant, in order to improve the methods in view of achieving this.
No, it would be a waste of resources to set up an ISO ring test at the present state.
From the studies reported so far, it appears that in-vitro migration rate of phthalates from a PVC product depends upon two factors, the matrix and the conditions employed for the forced migration. The matrix related factors - phthalate content, homogeneity of the phthalate distribution in the matrix, other matrix constituents - all seem to influence migration rate. The test conditions which have been shown to influence in-vitro forced migration of phthalates from PVC-products appear to be: mechanical force applied, temperature and the saliva composition. Besides the above mentioned factors, the physical properties of a PVC-product, as a result of 'moulding' and 'curing' processes used for the manufacture of PVC products, may also affect the migration rate of the plasticiser.
It is suggested that for the validation of a test method for phthalate migration from commercial products:
- The target migration rate for the test method should be 9m g/min from the reference discs used in the Dutch volunteer study. This will allow to identify the products which may not comply with the recommended limit values. The test method, prior to a validation exercise using ISO guidelines, should demonstrate that products with DINP migration rates below/above the recommended limit value can be discriminated.
- The test samples should be products of a known matrix with respect to manufacturing process and other matrix factors described above. Test samples should include products produced using different manufacturing process and with different phthalate contents. The reference material should be available upon request by national authorities.
- The simulation conditions used for the test method for in-vitro migration should in principle be as close as possible to in-vivo conditions. It is possible to change the extracted amount by changing at least three different parameters: the physical force, the temperature and the composition of the saliva simulant. In the LGC methods the two first were used to "calibrate" against the target values, but the repeatability within a laboratory and the reproducibility among laboratories were poor. It may be worth to try to use different saliva simulant composition to see whether that gives less variations. To match the in-vivo studies, an artificial saliva with some organic content should be preferred.
- The metal balls used as a source of mechanical force in the LGC study may tear off the PVC sample. This process will result in poor repeatability. In the LGC report, no mention has been made concerning the physical appearance of the test samples after the completion of the experiment. Moreover, the particles release due to aberration of PVC may also affect the test results. In the report from JRC Ispra (CSTEE/ 97/1 - Add. 170), a reference has been made to a home made 'gnawing machine', which may be a better option as a source of mechanical force for the in-vitro test method.
- The phthalate analysis should be performed by an analytical technique which can identify phthalates unequivocally.
- The variability of the test method should be as low as possible, ideally < 20% (Allowance of 20% tolerance in the results permits 40% deflection from the target value!)
Final recommendation
- For further work with migration studies, the use of a standardised reference material is recommended. Such a material should be of the same composition as the one used in the Dutch in vivo experiment.
The US in-vivo and in-vitro studies
1. How does the CSTEE interpret the additional results of in-vivo tests in the US (CSTEE/97/1-Add.163/C) and the relation between in-vivo and in-vitro tests.
In the US in-vivo test ten persons chewed/mouthed discs cut from a toy containing almost 43% DINP, and the saliva was collected and analysed for phthalates. The procedure was divided in four times fifteen minutes and the saliva samples from these periods were analysed separately. The results show variations for each subject between the four periods (up to a factor of 3) and a variation between subjects of almost an order of magnitude. The average of all results was 4.3 µg/10 cm2/min and the maximum value was 9.7 µg/10 cm2/min, with an extreme for one 15-minute period of 13.4 µg/10 cm2/min. These results support those obtained in the Dutch and Austrian studies used for the earlier opinion of the CSTEE.
The in-vitro migration of DINP was also tested in the US study. Different sizes of pistons were used to mechanically treat the test material in different saliva simulants. The material used in the in- vivo test gave a migration of 3.3 µg/10 cm2/min with extremes of 2.5 and 4.1, respectively, in a series of five determinations. The in-vitro method produced a mean close to the mean value from in-vivo study, but the repeatability seem to be rather low (few data points) and it is not possible to say anything about the reproducibility among different laboratories.