TECHNICAL ADVISOR'S REPORT TO THE FOOD, DRUG, AND COSMETIC PACKAGING MATERIALS COMMITTEE

June 18, 2001

Lester Borodinsky, Ph.D.

Ladies and Gentlemen:

It is a great pleasure for me to be with you once again in the role of the Committee’s Technical Advisor. This report will describe one of the current technical issues we are dealing with in establishing satisfactory regulatory positions for plastic food-contact materials. Specifically, this report will provide observations on exaggerations inherent in estimating dietary exposure to components of food-contact articles. While not directly a consequence of the Food and Drug Administration’s (FDA) implementation of the Agency’s Food-Contact Notification (FCN) program, certain new requirements installed in the new program focus attention on the need for accurate estimation of dietary exposure.

Food-Contact Notifications

As we have been reporting to you, FDA officially set its FCN program into motion on January 18, 2000. As you will recall, FDA issued draft guidance documents regarding the chemistry and toxicology needs for Notifications entitled “Preparation of Premarket Notifications for Food Contact Substances: Chemistry Recommendations” and “Preparation of Premarket Notifications for Food Contact Substances: Toxicology Recommendations,” respectively, in September 1999. In addition, FDA promulgated a Proposed Rule on July 13, 2000 (65 Fed. Reg. 43269) intended to formally implement the FCN process, along with a revised companion administrative procedures guideline entitled “Preparation of Premarket Notifications for Food Contact Substances: Administrative.” Despite the absence of such implementing regulations and final versions of the guidance documents, the program is “up and running.” Although the official guidance documents and regulations are not formally in place, the procedures and parameters of the system are beginning to take shape on a case-by-case basis. In fact, numerous notifications already have become effective.

One of the information requirements not previously required of petitioners for food additive petitions that now has been “institutionalized” with the implementation of the FCN program is the need for cumulative estimated dietary intake (CEDI) values. In the review of food additive petitions, the petitioner has always been required to provide information from which to calculate an estimated dietary intake (EDI) for the proposed use of the substance. In conjunction with the EDI thus calculated, FDA’s reviewers of the petition often also considered all prior clearances of the particular substance (if any) along with the EDIs for the prior cleared uses - the total of the EDI for the proposed use and the EDIs for the existing uses is referred to as the CEDI. Under the FCN program, the guidance documents indicate that the applicant needs not only to calculate the EDI for the notified use but also to include the CEDI. The change in this instance is that the Notifier, not FDA, is responsible for the CEDI. This added requirement to include the CEDI in FCNs highlights the need for accurate estimates of exposure for individual uses. In essence, while an “overestimate” may be acceptable in a given application, such overestimates generally are added and can result in compounding of overestimation.

Estimating Dietary Exposure

FDA has set forth a means for assessing the dietary exposure to components of food-contact articles that may potentially migrate to the contacted food. The Agency’s model has been used since 1981 and it is an integral part of the “chemistry” guideline documents ever since that time. In the agency’s model for assessing dietary exposure, both the level of potential migration to the contacted food and the use patterns for the packaged food are employed.

FDA uses the term Consumption Factor (CF) to describe the portion of the diet likely to contact specific packaging materials. The CF is the ratio of the weight of food contacting a specific packaging material to the weight of all food packaged. FDA has established CF values for both packaging categories (e.g., metal, glass, polymer, and paper) and specific polymers. In addition, FDA has recently included refinements or subdivisions of certain polymers where the Agency has received adequate data upon which to base such a subdivision. FDA's current CF values are set forth in Appendix IV of the Chemistry Recommendations document.

In addition, FDA uses the term Food-Type Distribution Factor (fT) to reflect the observation that the “foods” with which particular polymers are in contact are different from each other and that different types of food behave differently with regard to migration. The types of food that FDA has recognized are aqueous, acidic, alcoholic, and fatty. FDA's current fT values also are set forth in Appendix IV of the Chemistry Recommendations document.

In essence, the CF and fT are FDA’s means of capturing the use patterns for the food-contact applications involving a particular polymer or type of application.

The other type of “variable” that is used to calculate an EDI is the results of migration studies; the experimental parameters used in the migration studies are designed to reflect the intended conditions of use. Once again, the Chemistry Recommendations document is instrumental in the design of the migration studies. While these guidelines are quite informative, they are not “recipes” for performing migration studies, but rather, are descriptions of test elements to be considered in designing a migration study. The critical factors to consider in designing such studies are summarized below:

Samples. The samples should reflect the highest use level of the material of interest and the sample used should be the maximum thickness anticipated. In this way, all lower use levels or lower thicknesses are covered.

Simulants (solvents). FDA has provided recommended food simulants for performing migration studies. For simulating aqueous, acidic, and low alcohol content (up to 15% alcohol) foods, 10% ethanol should be used. To simulate fatty foods, FDA recommends the use of an edible oil such as corn oil or olive oil; as an alternative, HB307, a synthetic triglyceride, or Miglyol 812, a fractionated coconut oil, may be used. If the use of a liquid fat poses analytical problems, an aqueous alcohol solution may be used as an alternative fatty food simulant. Depending on the polymer class, FDA recommends either 50% or 95% (v/v) aqueous ethanol as appropriate alternative fatty food simulants.

Time and temperature. Testing should be conducted under the most severe conditions of actual use. The testing conditions mirror the actual conditions, the most frequently referenced of which are those referred to as Conditions of Use A through H. For example, FDA’s recommendation for simulating Condition of use A (“High temperature heat sterilized, e.g., over 212EF”) is to perform the testing at 121EC for 2 hours followed by 40EC for 10 days. If less severe use conditions are of interest, different test conditions are recommended that are intended to reflect the actual use conditions.

Analysis of extracts. The extracts should be analyzed at the specified test periods for the specific potential migrant(s) of interest; the specific analytes must be determined based on the chemistry of the “new” substance for which the regulatory status is being established. While any suitable analytical procedure may be employed to determine whether the substance is extracted by the test solvent, the method should be reasonably specific to the substance and the test results must be validated in accordance with the procedures that FDA sets forth in its testing guidelines

Once the migration data are available, and the CF and fT values are known, these values are used together to calculate the EDI; a general expression to describe the calculation is as follows:

EDI = CF x <M> = CF x S(M_if_i)
= CF x [M_aqf_aq + M_acf_ac + M_alf_al +M_ff_f]

where CF is the consumption factor for the polymer (or application) of interest, M_aq, M_ac, M_al, and M_f are the migration results for the simulants representing aqueous, acidic, alcoholic, and fatty foods, respectively, and faq, fac, fal, and ff are the food-type distribution factors for aqueous acidic, alcoholic, and fatty foods, respectively, for the polymer (or application) of interest.

Using the above approach, EDIs are calculated. While it appears that the above approach yields accurate estimate of dietary exposure, there are exaggerations in numerous places that lead to inherently inflated EDI levels, as described below.

Exaggeration of EDI – Consumption Factor

In some instances, the consumption factors per se are intended to be, and are, reasonably accurate fractions of an individual’s diet that come into contact with the particular material. However, there are many instances in which the consumption factor employed is greater than the fraction of food that will be packaged in the material. For example, FDA recommends use of a default consumption factor (CF) of 0.05 (5%) where no CF for the substance has been established and where it is not feasible to acquire valid data upon which to base a CF independently. In addition, several of the CF values set by FDA have not been re-evaluated since they were established by FDA 20 years ago (such as uncoated or coated paper and paperboard) and we have a strong suspicion that the default CF values for these categories (0.1 (10%) and 0.2 (20%) for uncoated and coated paper, respectively) are significant overstatements of actual food-contact use. Thus, the CF value used will often lead to an overestimate of the EDI.

In addition, in using a default CF for evaluating the EDI for a polymer additive, the necessary assumption that is made is that the substance used in the polymer is used throughout the industry for that polymer, i.e., that there is 100% market penetration. While there are a few substances that are used quite widely (based on our experience in reviewing formulated products), these are few and far between. In most instances, the use of a given polymer additive competes with other substances which give a comparable technical effect. Thus, in using the full CF value for a given polymer, the 0.12 (12%) CF for low density polyethylene (LDPE), for example, one assumes that all LDPE will always use the substance. This assumption of total market penetration was arguably reasonable in the era when broad clearances via food additive regulations were obtained, as the clearance obtained permitted anyone to look to the regulation and use it as a basis for establishing a proper regulatory status, even if the company using the clearance was not the company that submitted the food additive petition that led to the regulation; however, the total market penetration notion was undoubtedly an exaggeration even for such broad regulatory clearances, at least in most instances. This concept is recognized, at least in part, in FDA’s “model” for evaluating colorants – FDA indicates that it is appropriate to use a CF of 0.05 (5%) to obtain a clearance for a given colorant for use in all polymers, even though the total CF for all polymers is theoretically approximately 0.8 (80%) (including all coatings). Our understanding of the reasoning behind this sensible approach is that FDA recognizes that all food-contact articles will not be blue, or orange, or purple, or pigmented at all. Thus, because of the ability to visibly distinguish between pigmented and unpigmented plastics, or between one pigment and another (at least based on color), FDA has permitted something less than total market penetration. In essence, FDA has allowed a “subdivision” to account for a technically limiting factor, i.e., color. While it is feasible to take into consideration technical limitations for other polymer additives where the technical effect of the substance will only be employed under certain circumstances, e.g., only for use in films and not in thick articles, this is a road often not taken, as it is frequently the case that the EDI calculated using the full CF value (i.e., assuming total market penetration) is low enough to be able to rely on the existing toxicology data to establish safety. In such instances, the sensible approach is to not “bother” to more accurately estimate exposure, as there will be no advantage gained for the additional effort (and expense) that will be needed to justify the apportionment of the full CF even where such a partitioning is completely justifiable. This, in fact, is the path most often taken in these instances and it necessarily leads to an exaggeration in the EDI. While the overstated EDI “causes no pain,” i.e., no additional data need be acquired, it is, nonetheless, an exaggeration of a “true” EDI.

While it was perhaps a reasonable approach for the broad clearances obtained in food additive regulations applicable to everyone, it is, in most instances, certainly not an accurate reflection of the EDI to assume total market penetration for the evaluation of FCNs, as these become effective only for the company submitting the Notification. In essence, the EDI based upon a default CF value determined in an FCN submitted by a company for a material can only be viewed in one way – the EDI thus calculated encompasses all other companies’ use of the material, even if none of them will ever obtain a clearance. It does allow FDA the easy pathway of looking to an EDI calculated using 100% market penetration for subsequent FCNs for the same substance used in the same application(s) – and the subsequent notifier, likewise, has an easy pathway – as FDA (and the subsequent notifiers) can look to the initial calculated EDI to likely encompass the “additional” uses; this works as long as FDA is mindful not to add the EDI from the “new” uses to the original EDI. Nonetheless, in this way, EDI values determined under the FCN process will most often be exaggerations.

Exaggeration of EDI – Migration Studies

As noted above, several parameters must taken into consideration in designing a migration study so that the test conditions cover the actual use conditions for the material for which the clearance is being sought. In each case, the testing parameter must be performed in a worst-case manner, as described below.

Samples

As noted above, the samples should reflect the highest use level of the material of interest and the sample must be the maximum thickness expected in actual use. In practice, the level of a polymer additive used in the test sample is far greater than the level(s) that will occur in actual use. It is well known that the level of migration will vary with the use level so that higher use levels generally give higher levels of migration; the migration level often will be directly proportional to the use level. In essence, the use of migration study results obtained solely using samples with the highest use level (and hence, the highest migration level) in calculating the EDI involves the assumption that the migration is always at the elevated level. In this way, the EDI will exaggerate actual exposure.

Likewise, use of thicker test samples will almost always give higher migration results than thinner ones, as the amount of the migrant in a given surface area exposed to the test solvent will be greater in the thicker article. Thus, a calculated EDI based on migration studies from test samples that are thicker than the actual food-contact articles will be larger than the “real” EDI.

Simulants

The food simulants used in migration testing are generally selected to ensure that the corresponding migration to food that is being simulated is not underestimated. In setting up such a simulation, therefore, the simulants themselves often necessarily exaggerate the levels that would be observed in food in actual use. There are a few publicly available reports of testing performed using both food and food simulants that illustrate that FDA’s recommended food simulants exaggerate the action of food on migration of components of food-contact articles.

FDA recognizes that the worst-case fatty food simulant generally is a liquid fat; FDA’s preferred simulant is corn oil, although other liquid fats are viewed as equivalents. Furthermore, for polystyrene, for example, FDA explicitly permits the use of 50% ethanol as an alternative to the use of a liquid fat, since it is often technically challenging to employ the latter for gravimetric or even instrumental analyses. Much of the experimental work on alternative fatty food simulants for polystyrene was performed by FDA. For example, in 1982, FDA published a paper summarizing studies involving water, 3% acetic acid, 8% ethanol, 30% ethanol, 50% ethanol, corn oil, and sunflower oil. Tests were performed at 40°C and employed polystyrene having three different levels of residual styrene. In these studies, FDA reported that the diffusion coefficient for 50% ethanol was about 17 times greater than that for corn oil; as migration is usually viewed as proportional to the square root of the diffusion coefficient, the migration to 50% ethanol is approximately 4 times that to corn oil. FDA also studied migration to actual foods: milk, cream, beef, beef fat, gelatin, margarine, mayonnaise, vanilla frosting and enrobing chocolate. The level of styrene migration to corn oil was greater than that for each of the foods, by factors ranging from 2 to 40.

In another study, FDA analyzed several solvents for styrene migration from polystyrene: water, 3% acetic acid, 8% ethanol, 20% ethanol, 50% ethanol, 100% ethanol, corn oil, HB-307 (a synthetic triglyceride), hexadecane, and decanol. The solvents that show the greatest levels of extraction were 50% ethanol and 100% ethanol. The styrene migration levels to 50% ethanol were approximately 3 times that for corn oil, corroborating the earlier FDA work.

These studies illustrate that 50% ethanol, which is often used as an alternative fatty food simulant for several polymers, including polystyrene, gives greater levels of migration than do edible oils; in turn, the edible oils yield greater migration levels than do actual fatty foods.

In another study, performed on behalf of FDA by Arthur D. Little, Inc., migration of two widely used antioxidants from low density polyethylene (LDPE), high density polyethylene (HDPE), and polypropylene to water, corn oil, low fat foods (baby food (banana), baked beans, evaporated skim milk, no-fat gravy, and apple jelly), emulsified fat foods (chicken a la king, evaporated milk, liquid nutrient, process cheese spread, and Newburg sauce), and free fat foods (baby food (turkey), beef stew, chicken broth, chicken soup, corned beef hash, and gravy). With regard to LDPE, the migration levels to low fat foods were less than to water, and the migration to corn oil was 3 to 8 times that to emulsified fat foods and 2 to 8 times that to free fat foods. With regard to HDPE, the migration levels to low fat foods were equal to or less than to water, and the migration to corn oil was 2 times that to emulsified fat foods and roughly equal that to free fat foods. With regard to polypropylene, the migration levels to low fat foods were equal to or less than to water, and the migration to corn oil was 2 to 25 times that to emulsified fat foods and 2 times that to free fat foods. This is another example of a study that indicates that migration to food simulants will, in most cases, exceed the migration to actual foods.

The foregoing are a few examples of experimental studies that show that the use of food simulants exaggerates migration to actual foods; there are undoubtedly other similar examples.

This expectation of the exaggerative nature of food simulants, particularly the use of edible oils as fatty food simulants, can be seen by turning to the Council of the European Communities Directive 85/711/EEC, “Laying down the list of simulants to be used for testing migration of constituents of plastic materials and articles intended to come into contact with food stuffs.” In the Directive, the Council indicates that, in certain instances, the migration results obtained using the fatty food simulant (olive oil or one of its acceptable alternatives or substitutes) should be divided by a “reduction factor” to account for “the greater extractive capacity of the simulant for such foodstuffs.” In fact, such migration testing for most types of food permit the application of a reduction factor. The only foods for which a reduction factor is not permitted are fruit preserves in an oily medium (food “type” reference number 04.02.C.II), preserved vegetables in an oily medium (04.05.C.II), animal and vegetable fats and oils (other than margarine and butter) (05.01), preserved meat and fish in an oily medium (06.05.B), and sauces containing oil and water forming two distinct layers (08.06.C). The simulation of all other fatty foods using fatty food simulants involves the reduction of the actual migration results by a factor ranging from 2 to 5 to account for the observation that migration to fatty food simulants exaggerates migration to most fatty foods.

For the above reasons, migration study results using fatty food simulants overstate the expected migration to most fatty foods.

Time and Temperature

It is well known that the most important factor in governing the level of migration is temperature. This is a critical consideration, because, as noted above, testing should be conducted in a manner that simulates the most severe conditions of actual use. In most instances, test results from the most severe test condition, reflecting the most severe actual use conditions, are the only data available in a given situation, i.e., migration testing is usually not performed under more than one set of conditions and that single data set is acquired using the most severe conditions of interest. Hence, the results obtained in such instances, which are intended to represent the level of migration under the most severe use condition, are applied to all uses of the material; these results, in turn, are converted to EDI values by application of the complete CF values, as discussed above. The result of this procedure is that, in effect, the results obtained under the most severe conditions (such as retort or hot fill conditions) are assumed to be the same as what would be expected under less severe conditions. The significance of this observation is that, for most materials, the milder actual use conditions are much more widely in use than are the severe use conditions. For example, this Committee’s Polyolefin Safety Task Group issued a report in May 1995, entitled “Report on Polyolefins Used in Food-Contact Applications;” the report was provided to FDA in Food Additive Master File (FMF) No. 582. In this report, the Task Group concluded that “the vast majority (96.3%) of the applications involve conditions of use at room temperature and below.” Consequently, the results of migration testing under conditions intended to simulate more severe use conditions, which is usually a small fraction of the overall use, is used to represent all conditions of use (as a worst case). In this way, the conventional way of performing migration testing (i.e., only under the most severe use conditions) leads to an exaggeration in calculating the EDI, as the relatively high levels of migration, obtained for the small minority of uses, are applied to all intended uses, which are usually room temperature (or below) use and for which a relatively low level of migration would otherwise be observed.

Analyses of Extracts

As noted above, extracts are analyzed at the specified test periods for the specific potential migrant(s) of interest. In some instances, however, the available analytical methodology does not permit a specific analysis but instead is general enough that the analyses could encompass not only the substance of interest but also other substances that could arise from cleared materials that are needed for fabricating the test sample. In such instances, the results reflect not the substance of interest, but also these “other” substances. In such instances, the migration results could potentially overstate migration of the specific substance and, in turn, the EDI calculated from these results would be an exaggeration.

Conclusion

The foregoing demonstrates that EDI values calculated in a way to represent a reasonably accurate dietary exposure are, almost without exception, exaggerations over actual dietary exposure levels. This would arise even there is only one instance of exaggeration in one of the elements used to estimate exposure. However, the foregoing demonstrates that exaggerations likely arise in numerous places for each EDI calculation; in such instances, the exaggeration is compounded.

We bring this to your attention not to advocate a change on the way EDIs are determined. In exaggerating the “true” exposure, we should, and do, derive comfort in appreciating that there are additional “margins of safety” in safety assessments, as safety evaluations are accomplished by comparing the calculated EDI value (exaggerated, as demonstrated above) to a level deemed to be safe, generally express as the acceptable daily intake (ADI). Hence, we prefer to continue to employ the conventional procedures, knowing that the added safety cushion exists. However, it is equally important to realize that the values are inflated over “reality,” particularly as we are prone to refer to such calculated EDI values as if they are true-life values. Therefore, we advise you to be mindful of the exaggerative nature of EDIs when discussing or describing exposure values. Also, where is feasible, it is worthwhile to remind the reader (or listener) of such descriptions of the concept that, while EDI values calculated in the conventional ways described in this report are safe, they are also exaggerations of “real” exposure.

More About SPI:   Vision and Mission . Membership . Business Units . Regional Offices . News and Publications . Calendar of Events . Terms and Conditions of Use