Priority-based assessment of food additives database of the U.S. Food and Drug Administration Center for Food Safety and Applied Nutrition.

The priority-based assessment of food additives (PAFA) is a database maintained by the U.S. Food and Drug Administration (FDA) Center for Food Safety and Applied Nutrition. PAFA contains extensive administrative, chemical, and toxicological information on 1685 regulated direct food additives. The database also has limited administrative and chemical information on an additional 1236 direct additives. The total 2921 substances represent everything added to food in the United States. PAFA contains up to 150 different kinds of information about each chemical. Administrative and chemical information includes Chemical Abstracts Service Registry numbers, Code of Federal Regulations citations, the annual usage and estimated daily U.S. human consumption, the Joint Committee on Food Additives Allowable Daily Intakes, the FDA Redbook structure categories of the chemicals, and their technical effects. Toxicology information shows the type of studies done for each chemical, the species of animals tested, the toxicological effects observed and the sites where they were seen, the lowest doses that cause a toxicological effect in each study, a source citation, and other types of related information.

The U.S. Food and Drug Administration (FDA) Center for Food Safety and Applied Nutrition (CFSAN) has the mandate to assure the safety ofthe U.S. food supply. To help accomplish this task, the Center has created a database that currently contains all oral toxicity information available on almost 1700 of approximately 3000 direct food additives regulated in the United States. This database is called the priority-based assessment of food additives (PAFA).

History of PAFA
The origin of PAFA can be traced to 1969 when, in the wake ofquestions concerning the artificial sweetener cyclamate, President Nixon asked that the toxicology ofall gererally recognized as safe (GRAS) compounds be reviewed. The scope ofthe project was expanded in 1977 by former FDA Acting Commissioner Gardner, who ordered periodic reviews of all oral toxicology studies of GRAS compounds as well as direct food additives. These cyclic reviews were to result in the creation of listings of food additives in order oftheir priority for the requirement offurther toxicological evaluation. The database resulting from these reviews was to be called the cyclic review ofingredient safety profiles (CRISP).
The mammoth task ofcompiling all the necessary information finally started in 1981 when toxicologists and consumer safety officers ofthe (then) FDA Bureau ofFoods began to create extensive, evaluated, and abstacted toxicology safety profiles for hundreds offood additives. At about this same time, when CFSAN managers realized that the size of the task compared to the resources available would not allow for the regular periodicity ofthe reviews, the name ofthe database was changed to the current priority-based assessment of food additives. However, as indicated by this name, the ability to compare the safety ofdifferent food additives was retained. During this same time, the electronic storage ofPAFA began, using the resources ofmainframe computers at the Federal government's Parklawn Computer Center in Rockville, Maryland, and the Model 204 database programming language.
Upon further realization ofthe resources this task demanded, in 1987 an outside contract was awarded for reviewing, updating, and further expanding the PAFA database. This contract was renewed in 1990 for continuing the updating ofPAFA as well as enlarging the database to include all genetic toxicology studies of substances previously reviewed and all toxicity data for the remaining 1300 (mostly GRAS) compounds, which up to this time could not be included.
Another major event came in 1989 when the computer software used to access PAFA was reorganized and simplified, and additional data fields were added. This improvement is now leading to the next major milestone, when in 1991 the PAFA databasewillbemadeavailabletothegeneralpublicviadirecton-lineaccessthroughaccountswiththeU.S. NationalTechnicalInformation Service, the database searching capability ofthe National Library ofMedicine, and through general publication on compact disk (CD).
Other plans for the near future include entering all PAFA genotoxicity data into the U.S. Environmental Protection Agency Genetic Activity Profiles (GAP) database and modifying the software that is used to operate GAP to display in similar format the general toxicological data in PAFA.

Information in PAFA
The PAFA projectinvolves manypeople fromdifferentdivisions in FDA/CFSANandthecontractgroupengagedincontinually updatingandmaintainingthedatabase. Administrativeandchemical information concerning the compounds in PAFA is reviewed regularly. Forexample, thepoundageofeachdirectfoodadditive annually used in the U.S food supply is monitored periodically through contracts with theU.S. National Academy of Sciences. In general, to make the most efficient use oflimited resources, the greatertheamountofthe chemicalthatisused, themorefequently the toxicology information is updated: All oral and genetic toxicity data for foodadditives usedatlevels exceeding 1000pounds (460 kg) each yearareupdatedatleast annually; thetoxicology dataon chemicals thatarenotcurrentlybeingusedatall arenotupdated.
Paper copies ofall primary data collected for the PAFA database and their written reviews and evaluations are kept at CFSAN in a consistent format in hard-copy"food additive safety profile" (FASP) folders for each individual compound or group ofcompounds. All information is independently checked twice for accuracy; then most of it is also stored in the computer database. The Center has developed computer programs to access this information and to rank the compounds in order of concern to determine those chemicals that need additional attention and the toxicological consequences of substituting one food additive for another. The criteria for choosing the toxicological evaluation and ranking were first published by FDA/Bureau of Foods in 1982 in Toxicological Principles for the Safety Assessment of Direct Food Additives and Color Additives Used in Food (1), commonly referred to as "the Redbook." PAFA's primary function is to serve as CFSAN's institutional memory for the toxicological effects ofdirect food additives. The database is also routinely used by CFSAN managers and review scientists to help evaluate the overall safety ofthe additives that have been reviewed and entered in a standard abstracted format in the PAFA database, to find compounds with certain defined toxicological effects, and to rank compounds for their relative toxicity in comparison to other food additives.PAFA is also used by CFSAN consumer safety officers as an easily accessible and consistent source of information to answer outside inquiries, by CFSAN laboratory scientists for background information for research they are doing or planning, and by various CFSAN professionals as a source ofinformation for publications (e.g., [2][3][4][5]. PAFA is composed of an umbrella file called EAFUS (everything added to food in the United States) containing all compounds known to be added to food in the U.S. food supply, including direct, secondary direct, and GRAS compounds. PAFA does not contain substances that are indirect food additives, food contaminants, or natural food constituents, but their future inclusion is desirable. All compounds in the database have at least a minimum assemblage of information, including the Chemical Abstract Services (CAS) Registry numbers, the U.S. Code of Federal Regulations (CFR) citation numbers, and the best estimate ofthe U.S. population exposure (ifavailable). Well over halfthe compounds have complete, up-to-date toxicological information as well. As of May 2, 1991, toxicological information was available on 1685 regulated direct food additives and minimal information was available on 1236 additional additives for a total of 2921 substances in PAFA.
PAFA contains up to 180 different kinds ofinformation (fields) about each chemical. Administrative and chemical information includes CAS numbers, the places where the chemkjls are listed in the US. Code ofFedeml Regulations, the pounM disappearing into the U.S. food supply annually, the estimated daily human consumption (milligram per kilogram body weight per day) for each chemical, the Joint Expert Committee on Food Additives (JECFA) allowable daily intakes, the Redbook structure categories of the chemicals, and their technical effects.
Toxicology information includes the type of studies done for each chemical, the species ofanimals tested, the toxicological effects observed and the sites where they were seen, the lowest doses to cause a toxicological effect in each study, a literature citation for the studies ifpublished or a CFSAN Document Control Center reference ifreports are unpublished, and other types of related information.

Specific Structure and Contents of PAFA
To aid in searches of the database, PAFA's electronic storage is divided into two major tes ofrecords: administrative/chemical (RECITYPE=1) and toxicological (RECIYPE=2). These are then further subdivided into smaller sections called BOXes. Although RECTYPE=1 and BOX=1 are synonymous, REC-TYPE=2 is divided into eight BOXes. BOX=3 contains the genotoxicity data, BOX=7 holds acute oral toxicity information, and BOX=9 (currently containing the vast majority ofthe studies in the database) has all oral toxicity data other than acute studies.
The information in the other five BOXes in RECTYPE=2 are derived from that in BOX=9 to facilitate access. BOX=4A contains the data from the one reliable study in BOX =9 that demonstrates the lowest effect level (LEL) in a mouse or rat for the compound ofinterest; similarly, BOX=4B highlights the data from that one study in BOX=9 that demonstrates the LEL in dogs for the compound ofinterest, and BOX=4C shows the LEL in any species for the compound in question.
BOX=6, on the other hand, holds the data from a study in BOX=9 that shows the highest no-effect level ofthe chemical in the same species as listed in BOX=4C. Finally, BOX=8 groups, for emphasis, those studies in BOX=9 that demonstate toxicological effects of high concern such as hyperplasia, teratogenicity, etc. The specific fields in the database are listed and defined in detail in the Appendix. PAFA continues to be a highly valued resource of FDA/ CFSAN. We are looking forward to making it available universally so that it may become a valued resource for all persons interested in the safety of food additives.