61474 Federal Register I Vol. 45, No. 181 I Tuesday, September 16,1980 I Notices
DEPARTMENT OF HEALTH AND Table of Contents-cyclamate-Gonlinued " Table of Contents-cyc'amato-Contlnucd HUMAN SERVICES 2. Inconclusive But Suggestive of a PositiveEffect b. Cattanach, at a!. (A-151) 3. Negative 0. Lorke.el aJ. (A-716) 4. DefICient d. Lorke.et at,(A-611.App. 19) Food and Drug Administration !y. Carcinogenicity: The Evfdence 6. DaficientStudies A. The Reviewof the TemporalY Commilleeof The Na a. BoneMarrowStud'.es [Docket No. 76F-0392] tionalCancerInstilule • (1) Collin(G-27) B. Inconclusive But SUggestive StudiesRalsingA Serious (2) Dick,et 01 (A-1m Question As To The Posslble Carcinogenicity of CycIa (3)Ford.et aI. (A-297) Cyclamate (Cyclamlc Acid, Calcium ,mate (4) Friedman, el nf. (A-195) Cyclamate, and Sodium Cyclamate), 1. The Occurrence of Lung and U"OI Tumors llIld (5) Khera, et aI. (A-222) Commissioner's Decision Lym pbesarccmas irrMice (6) Osor,et al. (A-274) a. Rudali,et aJ. (G-43) b. Bloodcell studiElS (animals) b. Brantom, at aJ. (G-43) (1) Mostardl, at aI. (A-264) AGENCY: FOQd and Drug Administration. 0. Kroes, et aJ. (G-76; A-734) (2) LlskarandCobo(A-241) d. Hardy, at aJ. (A-690) 0. Bloodcells studlElS (humans) ACTION: Final decision following a e. The SigM'lCaIlce of Lymphosarcomas (1) Dick, Ellal. (A-1m formal evidentiary public hearing. 2. The Occurrence of Bladder TUIllOfS In D~ect (2) Coulson (A-703) FeedIngStudlesin Rats d. Spermcell studies SUMMARY: The Commissioner of Food a. The Occurrence of Bladder Tumors In Sprague (1) Ford (A-29n. Frfedman (A-195),and OsCI (A Dawleyand WJSlar Rats 274) StudiElS and Drugs is issuing his Final Decision (1) Hicks, at aJ. (Oirect FeedingStudy)(G-2; A (2) Keziwara. Ell01. (A-21n concerning the food additive petition for 832) (3)Leonardand Unden(A-240) ), (2) Ikeda,at aJ. (G-79) 7. Additional Support In Vi'1rO Cyt~nlo StudiD(J Sug~Slio;o the artificial sweetener cyclamate. The (3) Taylor.et aJ.(G-13) G. Other Studies InsuflJclent to OUIwelgh SlI.'d:es Commissioner has determined that : (4) Homberger.et aJ. (A-348) (5) 5chmaehl,et al (A-555) 1. Summary cyclamate has not been shown to be (6) Plank, at al (A-401-404) 2. Host·Me
To implementthe Food Additive also endorsed a prohibition,based on probable consumptionpatterns. safe Amendment,FDAcompiledan advisory safety grounds,of cyclamate in conditions of use of cyclamate can be "GRAS list" of substances already on beverages for general use and in the prescribed. the market. The final list of November future processing of general purpose The parties in the hearing were the 20.1959, included cyclamate.In 1961, food lid. at 13645). Bureau of Foods of the Food and Drug FDAadvised Abbott that sodium B. Administrative Proceedings Administration ["Bureau")and Abbott. cyclamate was no longer considered to See 21 CFR 10.3[a).::lThe Bureau . be a drug,and was considered to be On November15,1973, Abbott filed a contended that Abbott's food additive generallyrecognizedas safe as a food food additive petition (FAP 4A 2975) petition for cyclamate should be denied. ingredient. pursuant to Section 409[b) of the act Abbott, of course, contended that its In the early 1950's combinationsof seeking approval for the usc of cyclamic petition should be approved. cyclamate and saccharin gained wide acid. calciumcyclamate and sodium Testimony concerningthe issues in use in fabricated foods.To determine cyclamate [hereinafter collectivcly the hearing was submitted in written whether a mixture of cyclamate and referred to as "cyclamate") :s as form.Oral cross-examinationwas saccharin gave results differentfrom sweetening agents in food and for completed and-briefs submitted to the those reported in earlier experiments technological purposes 4 in food.It is this ALJ by January 23,1978. where cyclamate or saccharin was petition which is the subject of this On August4, 1978, the Administrative tested alone, Abbott, in 1967, contracted proceeding.FDApublished a notice of Law Judgeissued an Initial Decisionin with the Food and DrugResearch filing of Abbott's petition in the Federal which he found that cyclamate has not Laboratories,a private research Registerof February 8,1974 [39 FR4935). been shown to be safe. Specifically,the After reviewingAbbott's food additive institution, to conduct a study. In this ALI'found[ID at 38-39): 6 study, eight of 60 rats fed a 10:1 mixture petition to determine whether it met the (1) Cyclamatehas not been shown to .of sodiumcyclamate and sodium criteria for approval of such a petition set forth in Section409[c) of the act, the be safe as required by Section409of the saccharin for two years developed Federal Food.Drug,and Cosmetic Act bladder tumors.(Fora further discussion then Commissioner A. M. Schmidt [Zl U.S.C. 348). of this study, see SectionIV.E.3.c ["Oser concluded that the supportingdata did not establish that cyclamate is safe for (2) It has not been shown to a study"),below.) reasonable certainty that cyclamate Because the Food and DrugResearch its intended use. The food additive petition was therefore denied by ordcr does not cause cancer in man or Laboratories study implicatedcyclamate animals. as a possible carcinogen, the then in the Federal Registerof October 4, Commissioner of Food and Drugs, 1976[41 FR43754). (3) It has not been sho...'I! to a Herbert 1. Ley, removed calcium Abbott and the Calorie Control reasonable certainty that cyclamate is cyclamate,magnesiumcyclamate, Council. an industry trade group,filed not a mutagen. potassium cyclamate and sodium objections to, and a.request for hearing (4) In the event that the cyclamate from the GRAS list [then 21 on..the October 4,1976order; only carcinogenicityand mutagenicity CFR121.101) and limitedthe marketing Abbott, however, made particularized questions are resolved, the record in this of those cyclamate compoundsto objections.In the Federal Register of proceedingwould support a findingthat therapeutic uses as drugs[34FR17063. March 4,1977[42FR12515), the then the acceptable daily intake is five mg October 21, 1969). On August 27, 1970, ActingCommissioner, Sherwin Gardner, cyclamate/kg body weight/day or less. FDAconcludedthat there was no granted Abbott's request for a hearing (5) Even ifthe carcinogenicityand substantial evidence of effectivenessof pursuant to Section409[f) of the act. mutagenicityquestions were to be cyclamate compoundsat any level for The formal evidentiary hearing began subsequently resolved, the record in this treatment of obese patients and with a prehearing conferenceheld on proceeding does not establisli probable individuals with diabetes and therefore April 20,1977. The issues considered at consumptionpatterns of cyclamate to prohibited continued sale of cyclamate the hearing,as set forth b,}' the the extent necessary to establish safe containingproducts with drug labeling AdministrativeLawJudgeat the conditions of use. [35 FR13644). This action wits based on Prehearing Conference, were as follows: On June 26,1979, the then the advice of a MedicalAdvisory Group (1)Whether the evidentiary record Commissioner, Donald Kennedy,issued established by the Assistant Secretary establishes to a reasonable certainty an interlocutory order remanding the for Health and ScientificAffairs. that cyclamate does not induce cancer case to the ALJ to develop the evidence . Department ofHealth, Educationand when ingested by man or animals. further on certain issues relating to the Welfare.The MedicalAdvisory Group (2) Whether the evidentiary record safety of cyclamate.This order was establishes to a reasonable certainty published in the Federal Register of Footnotes continued from last page that cyclamate does not cause genetic August14,1979,with minor non from the definitionof "food additive:" The term damage and is not mutagenic. substantive changes [44FR47620). "food additive" means any substance the intended (3)Apart from the issues in Numbers1 use of whichresults or may reasonably be expected to result, directly or indirectly.in its becominga and 2 above. what does the evidentiary 'Dr. Michael S\Oed3. the discoverer of cyclamate, componentor otherwise affecting the record show as an acceptable daily abo appeared as a non-party participant (iil. His characteristics of any food [includingany substance intake level for cyclamate? appearance was 5'JbsequElllly strickenfo;: failure to intended for use in producing. manufacturing, (4) Whether apart from the issues in participale. See Z1 CFR1Z.45~e). packing,processing.preparing.treating,packaging, 'The followingabbre\iatioll3 have been used in transporting,or holdingfood:and includingany Numbers1 and 2 above. because of the clUng materialin the record: lnllial De:isiiln;ill; source of radiation intended for any such use),if Transcript:Tr.;Brim to the ALI:B:ief;E:
Petition for Permanent Listing; Final (Senate Report No. 2422,reprinted in the evidence carefully, conducts a fair Decision. 45 FR 6252 ijanuary 25,1980): [19581 u.s. CodeCong. and Admin. evaluation of the evidence, states its Benylin: Denial of Approval of News5301.) reasons for crediting or not crediting a Supplemental New Drug Apptication; FDA's interpretation of the term piece of evidence, weighs all the Final Decision: 44 FR 51512 (Angust 31, "safe" used in section!09 of the act is evidence, applies the correct statutory 1979). See 5 U.S.C. 556{d): 21 CPR consisfent with the act's legislative standards, and decides. 12.87(d). history. FDA's regulations provide that a As the discussion inSectionIV below In determining whether petitioner has food additive is "safe" if"there is a demonstrates. the evidence submitted in met Its burden. the agency must, as a reasonable certainty in the minds of this proceeding does not provide a logical matter, arrive at one of three competent scientists that the substance reasonable certainty of no harm from possible conclusions. First, it may find is not harmful under the intended cyclamate. Manyofthe studies contain that the evidence establishes that the conditions of use" 21 CPR 170~3(i). deficiencies and are, therefore, simply additive is "safe." Second, the agency Taken as a whole. then. Section 409 inadequate, whether to prove safety or may find that the 'evidence establishes means that Abbott has the burden of lack of safety. Of the studies in the that the additive is unsafe. Third, the proving that the data in the record record entitled to weight. a significant agency may find that the evidence is establish that there is a reasonable number suggest. though they do not such that the safety of the additive is certainty ofno harm from use of prove, that cyclamate is a carcinogen unknown or uncertain.By allocating the cyclamate. There is considerable _ and a mutagen-As a scientiflcmatter, , burden of proof to the petitioner. Section disagreement, however, about how to one can imagine studies which would 409 authorizes FDA approval of a food apply that principle. negate these suggestions of additive petition only in the first Abbott contends that the Bureau's carcinogenicity and mutagenicity. But no situation. Confronted with either the witnesses did not base their opinions on such studies are included in the record. second or third situation. the agency presently accepted scientific methods Those studies in the record in which no must deny the petition. . and that therefore the Bureau is carcinogenic or mutagenic effect was Although the term "safe" is not advocating a standard of "emotional found are either too insensitive to rely defined in Section 409 of the act, the certainty" rather than "reasonable on as proof ofsafety or do not detract legislative history of the Food Additives certainty" (Abbott's Brief at 2-7). I sufficiently from the studies which Amendment of1958 makes clear that the recognize that Congress did not intend suggest that cyclamate is a carcinogen term "safe" was not intended to require to impose II burden higher than or mutagen. In these circumstances. the absolute proof of safety. The House "reasonable certainty:' At the same petition must fan. for the evidence Report states that: time. it must be understood that what supporting it does not establish the must be proved to a reasonable safety ofcyclamafe. '" '" '"Safety requires proof of.a reasonable certainty is "no harm:' That burden may certainty that no harm wi11 result from the be hard to meet, for credible proof of ID. Carcinogenicity:The Scieo!i5c proposed use of an additive. It does not-and Framework cannot-require proofbeyond any possible some harm will undercut efforts (0 prove doubt that no harm will result under any no harm, even ifthere is not enough A. Criteria for the Emluation of conceivable circumstance. proof to make out a certain case of Carcinogenidty Studies . This was emphasized particularly by the harm. That is the way Congress Beginning in SectionIV. I examine the scientific panel which testifiedbefore the intended it, and for good reason. The carcinogenicity studies contained in the subcommittee. The scientists pointed. out that Food Additives Amendment of 1958 , it is impossible in the present state of food additive petition for cyclamate. protects against carcinogens, mutagens, Two major issues recur in that scientific knowledge to establish with and other dangers in our food supply. By complete certainty the absolute harmlessnesa discussion. One is "statistical of any chemical substance. allocating the burden of proof as it did, significance:" The other is "biological Congress asked FDA to be conservative Significance:" These fwo concepts are [H.R. Rept. 2284,85th Cong.•2d Sess., pp. In deciding whether to approve food 4-5. 1958.)(Emphasis added.) addltives.s applied to interpret the results of animal The Senate Report agreed with the Abbott also contends that. for a studies in which one or more groups of assessment of the term "safety" scientist to conclude that cyclamate has animals • are fed a test substance and contained in the House Report, noting: not been shown to be safe, there must "SJIh p:utIes re!}'1l11 thelr interprelatio-.-S of " " * Conscious of the fact that any be an "objective basis for the evaluation results from tests cendccted on laboratory animals, substance 01:'.for that matter, anyperticular of the data presented" (Abbott's Brief at Ideed. one ofthe mdm}irIgpre:niRs of this food known to be good for the health of 2-7). I agree. procrdir.g is Imtrer.l!ts !.-om such tes'-s embe human beings can be deleterious to the It is not possible, however, to provide used us a bastsfa: cletennicingthe safelyor health of an individual who insists Ol'l c;uclnc~~cpotm~of a test s~os!ano:ein a formula specifying precisely the humans••princip!e gene!a!Iyrecogniz2d by consuming inordinate amounts of it, the quantity and quality of evidence an committee agrees with the Food and Drug sclenU5!s.nus prinople was expreuly recognized Administration that. instead of insisting on applicant is required to submit in order In lKtlon 4Oll{u:[3J(A) of the act (the De!mey Cl.~)which CllllU:!13llis the denial ofII food proofbeyond any possible doubt that no to meet its burden. But the lack of a precise formula does not mean that the llddltive petition ifthe food additive in ~estion ••• harm will result under any conceivable Is row:;d to Indcce cancer whimln6es!edby man or circumstances from the use of a particular process lacks objectivity. Nor does a animal or ifit Is found,after tests which are additive '" '" '" the test which should lack of certainty mean a lack Qf uppfClpr.ate Cortheevaluation oftl!esafely offood determine whether or not a particular objectivity, especially where the subject a:ldlt,,\·e, to indu!% cancer in man or ammal.. •• additive may be used in a specific percentage matter is complex and the science (emph:ls:s addeJJ. 21 USc. 34a{eJ£3HM. That the De~an!!y Clause is no! being In\·oUd in ~ of relationship to the volume of the product to evolving. The requirement of objectlvity which it might be added should be that of proceeLllg does not preclude refel'1!nce to i! for is met, I believe, if the agency reviews Co~onaIln!ent reasonable certainty in the minds of purposes of ascr:tainillg with respect to use ofanimal data. competent scientists that the additive is not Courts have consutently up~eldp-e=ent harmful to man or animal. subject to the lIn an}' event, as discussedin Srelions IVand V below. the Bureau·s\\'1tn6~S did not hold the repIator)' adions against carcinogens orsuspected procedural safeguards provided in the bill evidence in thil proceedingto • ltandard hlgbr carclLo~-sbased, at least in part, oa J'e3U!ts from which assure the right to hearing and judicial than "l'1!asonllble crrtainly." but rethu evaluated It tes!5 onlabo=atOf)' animals. E.rl~~!!!!faJ Defer~ review. In IJght of presently accepted IclcnlJflC methods. Footnotes continued on next page 61478 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices
one or more control groups are fed the 100 (97percent). 10 Ifthe numberof that any resulting throat cancers are duo same diet and handled in the same. tumors found in treated animals were to the substance (rather than to the manner as the treated groups except four instead of five, and none was found irritation). that the controls do not receive the test in controls, the P-vaiue would be P=.061 The relationship between increasing substance...The incidence of tumors in (rather than P=.03) and there would dose of the test substance and the effCllt the treated group is then compared to thus be a greater likelihood that the observed is known as the dose response the incidence of tumors in the control result was due to chance (6.1 percent .relationship. Dose response relationship group. A finding that a test result has •rather than 3.0percent). In contrast, if is another consideration involved in u "statistical significance" involves the ten tumors' instead of five were found in determination of biological slgniflcance, use of statistical methodology to ' treated animals and none in controls, . Carcinogens are known to exhibit doso determine the probability that the the P-value would be P=.OOl (rather response relationships. 11 The presence observed difference, ifany, in the than P=.03) and there would thus be . of a dose relationship is looked for in incidence of tumors in treated animals less likeliliood that the result was due to studies employing more than ono doso compared to controls is associated with chance (0.1percent rather than 3.0 level of the test substance. Ifthe effect the test substance rather than a chance percent). observed increases as the dose lovol of occurrence. A finding that a test result 2.Biological Significance. "Biological the test substance increases, it is more has "biological significance" involves. significance," as its name implies, likely that the effect observed is duo to consideration of certain biological involves consideration of biological the test substance and more weight can factors which provide information about factors. Some of the factors typically be given to the results of the study. the proper interpretation of the results. considered are the methodology of the Conversely, the absence of a dose Together, these two criteria help study involved, the existence of a dose response relationship in studles whoro scientists to decide what, ifany, response relationship, the rarity of such a relationship would be expected conclusions can be drawn from the tumors, and the presence of similar to occur, may detract from the weight to results of a study. results in other studies (G-139 at 5). be given a study. For example, there may be an Another consideration involved in a 1.Statistical$ignificance. The term observed difference in tumors between determination of biological slgnlflcancu "statistical significance" is generally treated and control animals. Ifit is is whether or not the same effect occurs understood to refer to a conclusionthat determined. however, that due to a in more than one study. Ifit does, the there is a small probability that the mistake those treated animals with significance of the studies may be observed difference between control tumors did not receive the.test enhanced. and treated animals is due to chance. substance, then, obviously, the tumor As noted above, there is an This probability is expressed-as a difference in the experiment cannot be . interrelationship between statistical decimal, eg; P=.l. The smaller the p. attributed to the test substance. significance. and biological slgnificanco, value, the less the probability that the Similarly, if there is no difference in Scientists view the statistical and the effect is associated with chance and tumor incidence between treated and biological data together to determino hence the greater the likelihood that the control animals, but there is a what, ifany, conclusions can be drawn effect is associated with treatment. The substantial defect in the design or from the results of the study. larger the P-value, the greater the conduct of the study, the results of the It should be emphasized, however, • probability that the result is due to study would be considered biologically that neither statistical significance nor chance and hence the less the likelihood insignificant. biological significance supplies that the effect is associated with . The methodology of a study includes formulaic answers to questions about treatment. consideration offactors such as whether the meaning of data. They are very For example, assume that a study is animals in the study are randonily useful tools-analogous to canons of performed in which both treated and allocated to treated and control groups, statutory construction in assessing legal control groups consist of 100 animals whether treated and control animals are problems-but that is all they are. TheY' and five tumors are found in treated' handled in the same way, whether all must be used. as Commissioner animals and none in controls. In this control animals receive the same feed, Kennedy has said, with "the purposes of hypothetical study, the probability (P) whether all treated animals receive the the scientific enterprise" for which they that the observed difference in tumor same test substance, and the manner in . are being applied in mind (44FR 47622). incidence between treated and control which the test substance is 3.Position ofthe Parties andFindings. animals is due to chance is P=.03. A p administered. Each of these factors can ofthe ALJon StatisticalSignificallce. value of .03 means that the probability have an effect on the outcome, and must Abbott equates statisicalsignificance of the observed difference in tumor be considered in deciding how much with P<.05. In other words, Abbott incidence being due to chance alone is 3 weight to give a study. Suppose, for contends that only when the P-value for in 100 (3 percent) and therefore the example, that treated animals are the incidence of cancer in cyclamate probability of the observed difference administered the test substance through treated animals is less than or equal to being associated witli treatment is 97 in a tube which irritates their throats. The .05 can a study be considered positive better practice would be to insert the and therefore serve as a basis for Footnotes continued from last page same tube in the control animals, so that denying approval of a food additive Fundv.E.P.A., 598 F.2d 62. 87-89 (D.C. Cir. 1978J; their throats are subjected to the same petition (Abbott's Remand Brief at 15). EnvironmentalDefenseFundv, E.P.A•• 548 F.2d 998. irritation as the treated animals. Ifthis 1008-10 (D.C. Cir. 1976J. rehearing denied,548 F.zd "Lowering the dosage of carcinogens known to 1012 (D.C. Cir.1977J; Synthetic Organic Chemical is not done, one cannot be as sure as the statistical significance might suggest follow a dose response curvo can rosult In a Manufacturors Ass'll. v, Brennan, 506 F.2d 385. 387 "noncarclnogonlc" effect. Le;a dosago at which tho (3d Cir. 1974J. cert.denied.420 U.S.973 (1975J. This carcinogen wll1 not produce a statistically principle is also recognized throughout the record lOInsteadof using the decimal which expresses significant Increase In tumors (sooe.g. Soctlon (see,e.g., G-97at 1; A-647at 3-8J. _ the likelihood that the effect is due to chance (here, IV.B.3.b.(3J below], Ills Important to note, however, It should also be noted that use of animal data .03J. some statisticians refer to a confidence level that such a "noncarcinogenic" dosage of n kriown serves an Important ethical purpose as well: it that the effect Is due10 the treatment [here, 97%). carcinogen would not be considered safe because obviates the need for routine testing in humans of The two expressions are different ways of saying thresholds for carcinogens havo not been potential carcinogens. the same thing. • establlshed (see Tr. at 1068-69). . Federal Register I Vol. 45. No. 181 I Tuesday. September 16, 1980 I Notices 61479
Abbott contends that where the P-value placed on reports of positive results becalm! Traditional usage of a scientific method for the increased incidence of cancer in they are used to construct new hypotheses is not necessarily, however, a valid cyclamate-treated animals compared to and theories and will be incorporated into the reason for usage of that method in a body of assumed scientific knowledge. But no control animals is greater than .05, the - particular value of significance constitutes a particular case. study must be treated as negative and law of nature: it is a matter of scientific Moreover, although use of the P<;.05 therefore can provide a basis for custom, reflecting human value judgments as a standard is grounded in tradition. it approving a food additive petition about the purposes of the scientific Is no longer the method used by most (Abbott's Remand Brief at 18). enterprise. And in some contexts we are statisticians. Most statisticians. with the Abbott argues that use of the .05 especially troubled by the prospect of use of computers. ncw can and do report confidence level is standard and is mistakenly declaring that the result. of a to the precise P-value for an observed study are negative, Le., of mistskenly result and allow toxicologists and other supported by traditional usage (Abbott's saf~t)'. Remand Brief at 14). Abbott further concluding that a study demonstrates scientists to make a judgment for Such a decision, if incorrect. could result in contends that if carcinogenic effects that the widespread marketing of a carcinogen. A themselves on whether or not the level are nofsignificant at P<;.05 are used to regulatory agency may therefore have less of statistical significance obtained is conclude that cyclamate is potentially a reason than scientists do to insist on a very sufficient for them to reach a conclusion weak carcinogen, "science is done a high degree of certainty before concluding that the effect seen is the real effect of disservice and any hearing is an that a study is positive. Similarly. there ~ay the substance tested (G-139 at 4: see exercise in futility" (Abbott's Exceptions be reason for a regulatory agency to require also G-140 at 13). at 9).In Abbott's view, consideration of greater stringency than other lcientlsts In deciding how to apply the concepts require before concluding that a study Is of statistical and biological significance any carcinogenic effect that is not negative. statistically signifiance at P<;.05 as a In proceedings under Section 409 of the basis for concluding that cyclamate has (44 FR 47622; Bureau's Position Paper at act, we do well to keep in mind the fact. not been shown to be safe is a 7; see G-139 at 3-6.) . adverted to earlier. that evidence not "subjective and arbitrary treatment The Bureau further contends that the conclusive enough to confirm harm may [that] has never been the established strategy document Abbott relies on is yet be probative enough to harm to practice of the Agency" (Abbott's not the official position of the Bureau. is negate safety. Consider this example. Remand Brief at 14].In support of the not in evidence. and therefore should Suppose the data tell us there is a 90 out latter statement, Abbott relies on a not be considered (Bureau', Remand of 100 chance that canceris associated Bureau of Foods strategy document Reply at 4). The Bureau alsoasserts that with ingestion of the test substance (that which it claims shows that the Bureau the use of statistical criteria discussed in is P=.l). Ifthe rule of decision is that will not label a finding "positive" unless the strategy document is not we will not conclude that a substance that finding has a P-value of less than inconsistent with the position the causes cancer unless we think the .05 [Abbott's Remand Ex. at 23-25). Bureau has advocated in this proceeding chances are 95 out of 100 that it does • Abbott thus contends that the Bureau is (Bureau's Remand Reply at 5-6). The (ie., P=.05) then the data do not advocating in this proceeding a higher ALG adopted the Bureau's position that "prove" the substance is a carcinogen. standard than it ordinarily uses in effects which are not statistically But to say we lack proof of cancer is significant at P<;.05 may nevertheless reviewing food additive petitions. scarcely to say we have proof of safely. The Bureau recognizes that "out of support the conclusion that a food It is that distinction which is mandated convention P<;.05 continues to serve as a additive has not been shown to be safe by the statute. We are commanded to (IRD at 12-13). benchmark for statistical significance" seek proof of safely. not merely to 4. Commissioner's Findings on and that statistical significance at P<;.05 accept as proof of safety anything falling Statistical Significance. AlthoughP<..05 minutely short of proof of harm. may well be a prerequisite to labeling a has in the past been used as a standard. study unequivocally positive (Bureau's Commissioner Kennedy put it another this usogeis groundedin histolJ" not in way in pointing out that one's choice of Position Paper at 8; Bureau's Remand science (G-t39 at 4; A-859 at 3-4) or Reply at 5]. The Bureau contends, a P value may depend on the purpose to law. Beforethe advent ofcomputer which it will be put. In some cases, the however, that effects which are not technology, statisticians re?/edon. . statistically sjgnifi.cantat P<;.05 may consequences of a false positive are statistical tables to determinestatistical very serious. Suppose. for example, that nevertheless be relied upon as a basis significance(G-t39 at 4). These tables for denial of a food additive petition we are testing a new component for a generallyreportedonly three • rocket to be used in a moon shot, and (Bureau's Remand Reply at 5-6). significancelevels:.01, .os; and.t (id.). In support of its position, the Bureau that that component's survival is critical The use of P<;.05 as a reference point to success of the mission. In such a has adopted the following observations evolved from the use of these tables. made by Commissioner Kennedy in his circumstance, we would want to be Indeed, Abbott's witnesses seem to virtually 100% certain that the new Remand Order with respect to statistical recognize the lack of scientific basis for significance: component is more reliable than the use of P<;.05. One of these witnesses. Dr. component it is replacing. Thus, a P The use of "statistical significance" in the Smuckler, stated that "it is true that (the value of .000001 might be desirable. scientific community has not had the degree use of the .05 confidence level) is an Where, however, it is a false negative of inflexibility that the parties in these arbitrary decision, and, from a strictly that presents a problem, a test l'rith a P proceedings have assumed it has. Although mathematical standpoint. the selection value higher than.05 may supply the ".OS" confidence level has often been of this limit could be criticized. • used in the scientific literature to determine ,.u important information. In this whether a result is positive, th-ereis no fixed (A-859 at 4). Dr. Oser, anolher Abbott proceeding, there is good reason to be convention on the matter, * * * witness, could say only that the .05 seriously concerned about an incorrect confidence level is "commonly used" *** * * finding of safety, for the consequence is There is always a temptation to adopt the (A-858 at 24).Dr. Carlborg, a third the marketing of a carcinogen. Using this highest possible confidence level, particularly Abbott witness who is a statistician. did principle. there is a valid reaso?- f~r FDS in the scientific community where a very high not articulate any rationale for use of to consider effects tht are not SIgnificant value is given to the avoidance of a false P,.05, but rather stated that "NCI at P<.05 even though scientists or positive result. Especially high reliance is regularly uses the .05level" (A-857 at 9). regulators engaged in different 61480 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices
endeavors may not. 12 In so doing I the Bureau of Foods always uses P<.05 significant at P=.OO36. Finally, the dose emphasize that the difference between a as a standard. There is no evidence in response relationship between confidence level of P=.05 and P=.06 is . this record to that effect. Indeed, even cyclamate and the incidence of ' merely a matter of the degree of . the internal Bureau working paper lymphosarcomas in the Brantom study , certainty. In the former case, one is 95 which Abbott cites as support for its (discussed below) is statistically percent certain that the observed result position is to the contrary.13 The significant at P=.008. These studies is not due to chance. In the latter case, memorandum does state that the strongly suggest that cyclamate Is a one is 94 percent certain. There is no incidence.of a tumor should be carcinogen and therefore are suffiolent valid scientific rationale for concluding significant at P<.05 before a study will to raise a serious doubt concernIng the that there is a substantial difference be found to be positive (Abbott's carcinogenicity of cyclamate. Thus, even between these two confidence levels. In Remand Ex.; Exhibit 21 at 2). The if I were to use P'.05 as a standard. as the latter case, one is a little less certain memorandum further states, however, Abbott has suggested, I would about whether the carcinogenic effect is that "(i)f the data in a study indicate a nevertheless find that Abbott has fulled associated with treatment. I cannot, trend of increased tumor incidence that to show that cyclamate is safe. however, ignore such an effect. It may is not statistically significant at P<.05, _ 5. Position ofthe Parties, Findings of not be conclusive, but it is afleast doubts about the safety of the additive the ALIandCommissioner's Findings suggestive of a carcinogenic effect and will be raised which will warrant further On Biological Significance. Abbott therefore supports the conclusion that testing. This testing would in all . agrees that "evaluating effects for their the tested substance has not been probability require a chronic feeding biological significance. ifany, is a valld , shown to be safe. Such suggestive study with a 'higher power of test' e.g. scientific and regulatory exercise" results are especially important where more animals per group, higher doses (Abbott's Remand Brief at 15). they recur in a number of studies, for as etc," (id at 2-3). Thus, it is plain that the Moreover, it is undisputed that to a sclentlflc.matter, several inconclusive memorandum upon which Abbott relies, ' determine whether a tumor incIdence is but suggestive studies containing similar recognizes that effects which are not biologically significant, the results increase the likelihood that the statistically significant at P<.05 and consideration of biological factors, such effect observed is real (G-139 at 5: G therefore not conclusively positive, may as methodology of the study involved, 140 at 13). Adopting Abbott's suggested nevertheless raise a doubt as to the chemical structure, length of use, dOGO use of P<.05 for all studies would possible carconogenicity of a food response, rarity of tumors, and the preclude consideration of such additive. Itis therefore clear that the presence of similar results in other inconclusive but suggestive resultsand Bureau of Foods customarily considers studies is involved [Abbott's Remand therefore would be both scientifically effects that are not significant at P<.05 Brief at 16: G-139 at 5). The ALJ found and legally inappropriate. Ethyl Corp. v, where such effects raise uncertainty as that "biological significance must bo EPA, 541 F.2d 1,28 n, 58 (D.C. Cir.] (en to the safety of a food additive. attached to study findings where bane) cert. denied, 426 U.S. 941 (1976); Moreover, even ifthe Bureau had in borderline statistically significant Environmental Defense Fund v, EPA, the past used P<.05 as a standard, the effects occur (e.g. P=.06), but additional 598 F.2d 62, 89 (D.C. Cir. 1978): Color . Bureau's past practice is not controlling Ass~ factors exist" (IRD at 13). Mfg. v, Mathews, supra, 543 F.2d because the Bureau does not set the at 297. Abbott contends, however, that the agency's standards for approval of food , concept of biological significance can be I also reject Abbott's argument that in additive petitions. As the Court in evaluating other food additive petitions, applied only to reject effects that aro .Abbott Laboratories v, Harris, 79C 3732 statistically significant at P<.05 (N.D. decided June 12, 1980) made 12In my decision denYingapproval of a color m., (Abbott's Remand Brief at 15-15), but additive petition for Red No. 2.1 addressed an issue .clear, the function of the Bureau of cannot be applied to attribute similar to that raised by Abbott here. 1 emphasized' Foods' staff is to serve as advisors to the significance to effects that are not there. as 1do here. the Importance of using methods Commissioner (Slip Opinion at 3). The that are most likely to detect a carcinogenic effect statistically significant at P',05. I find. because of the consequences of mistakenly Commissioner makes all final decisions Abbott's "one way" test to be concluding that a food additive is safe:' and is in no way bound by the advice he untenable, for it would operate only to In reviewing the adequacy of the existing studles•. receives from the Bureau of Foods. prove safety, not to disprove it. 1have. in accordance with the philosophy of the Finally, it is important to note that, color additive law adopted a conservative approach Scientifically, it is just as appropriate to In order to be sure that the public health willbe although I find that it is appropriate to rely on biological factors to conclude adequately protected' • • 1have used methods rely on effects that are not significant at that an effect has biological significance. 'I. that are valid and arc also the ones most likely to the P<.05 level, I am not relying solely even though it is not statistically detect any carcinogenic effect that may be present, on such effectsin denying approval of ••• When a study is used to evaluate the safety of significant at P<.05, as it is to rely on a substance to be widely used by the public. the risk the food additive petition for cyclamate. biological factors to reject effects that of a false negative-of Incorrectly failing to detect The incidence of lung tumors in one are significant at P,.05 (G-139 at 4-6: an adverse effect that Is present-Is of greater strain of female mice in the Rudali study G-140 at 13). concern than the risk of a false positlve-6f incorrectly reporting an adverse effect when none (discussed below) is significant at Consideration of biological factors exists.' • • I am not. however. Imposing an P=.003 and the incidence of total can add further credence to or detract absolute standard of safety for evaluation of safety tumors in the same .strain of female mice from the weight that would normally be studies' • • I would not use a procedure. even ifIt and second strain of mice in the Rudali were the most conservative. ifthe procedure were given to findings with a particular p. not a valid one. If the questions about a substance study is also statistically significant at value. For example, two different types or the defects in a study are insubstantial. they do P<.05. Moreover, the incidence of of tumors may occur at the same P-valito not preclude approval of the substance. However. Iymphosarcomas in three combined when uncertainty remains about safety. aftet a fair in a particular study. Ifonly one of these evaluation of the record in accordance with generations of mice in the Kroes study tumor types recurs in other studIes, the sclenlific principles of evaluation. then. under (discus,sed below) are statistically recurring tumor type will be consIdered applicable law. the importance of protecting the to have greater biological significance public health must guide the final decision. FD&C IS Although the Bureau correctly notes that this Red No.2: Denial of Petition for Permanent Listing: memorandum is not in evidence and is not the than the tumor type that does not recur Final Decision: Docket No. 760-0033 Uanuary 25. • official position of the Bureau. 1have considered it in other similar studies. (The latter 1980.45 FR 6253J. because it helps to resolve this issue. tumor type may be found to be Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61481 insignificant it it does not recur in any where those decisions have been based positive effect). supports the conclusion studies.) on evidence that was inconclusive but that the test substance is safe (negative). Similarly, an effect may occur at P suggestive. In Ethyl Corp. v, EP.4, supra. or is inadequate for drawing any, value that. when viewed by itself, does the court stated thak conclusions as to the safety ofthe test not appear to be significant However, * * * we need not seek a singledisposlUve substance (deficient). These consideration of biological factors may study that fullysupportstheAdmInIstrators' classifications are discussed below. result in a conclusion that the effect has determination. Sciencedoes not worle that 1.Positive. A positive study is a study biological significance. For example, in a way:nor,for that matter,does adjudlcatory with contains results that establish that number of direct cyclamate feeding fact-fmdlng. Rather,the AdmInIstrator', a test substance causes cancer. Such a . studies in rats (see Section IV.B.2. decisionmay be Cully supportableifIt II study would result in a conclusion that below) more bladder tumors occurred in based, as it is, on the inconclusive but suggestive results of numerous Iludle••B,rIts the-food additive is unsafe under the cyclamate treated rats than occurred in nature. scientific evidenceI. cumulative: the general safety clause. and, under the controls. The occurrence of these tumors moresupporting, albeit inconclusive. Delaney clause of section 409 of the act, in each of the individual studies is not evidenceavailable.the morelikel)'the . would require that the food additive be statistically significant at P<.05. , accuracyof the conclusion. banned. There does not seem to be 'However, because bladder tumors are 541 F.2d at 37.15 much disagreement among the parties historically rare in the strains of animals The District of Columbla Circult Court concerning the definition of a study used in these studies, because the of Appeals recently reaffumed the which contains results which are occurence of these tumors in cyclamate opinion in Ethyl Corp. and further positive. Abbott contends that to be treated animals is consistent with a recognized that a regulatory agency positive a finding must be statistically small treatment effect, because the could not carry out its statutory significant at P<.05 and biologically occurence of these tumors in control mandate to protect the public from significant as well (Abbott's Remand R"<. animals is consistent with the incidence incompletely understood dangers such at 24:Abbott's Remand Brief at15). The of these tumors in historical controls, Bureau seems to agree with this and because these bladder tumors have as cancer if the agency could not rely on suggestive results: assessment (Bureau's Remand Reply at recurred in a number of studies 5: see Bureau's Position Paper at 2). * • * (R]egulations [prohlbiUng marketing of involving different strains of rats, these Although I agree that the level of bladder tumors are biologically a suspectedcarcinogen] mayjeopardize plants or wholeindustries. and thejob. statistical significance for determining significant. 14 dependlng on them.In such circumstances, that a study is conclusively positive To summarize, the concepts of the temptationto demandthat the agency should be at or near P=.05 and that the statistical significance and biological Curnish conclusive proofof carcInogenIcity u study should be biologically significant significance should be viewed together supportCor the regulations II great.However, as well. I am not deciding in this in determining the significance of a the decisionto delegateauthorityto an agencyto controlsuspectedcarcinogens is a proceeding whether the confidence level treatment related incidence of tumors. need be P=.05. Although the'Rudali, The closer the P-valve is to P<.05 the legislative judgment that is not open to Kroes and Brantom studies contain greater the confidence that can be questionin this court.Congress's directionto results that statistically significant placed in the results of the study. The EPAtoprotect against incompletely are factors to be considered in determining understooddangerscouldnot be carried out at well below P=.05 and suggest that ifwe were to adopt the proofrequirements cyclamate is a carcinogen, I find that. in biological significance may increase or allvocatedby industrY petltloners, decrease that confidence. This light of questions raised about the evaluation results in a decision as to EnvironmentalDefenseFundv. EPA, biological significance of these studies, how much, if any, weight a study should supra, 598F.2d at 89. Accord,ColorMfg. they are not conclusively positive (see B. below). In view of the fact be given (see G-139 at ~; G-140 at 13). Assn v.Mathews,supra. 543 F.2d at 297. Section IV Moreover, each study is not only See Herculesv, EPA, 598F.2d 91, 110 that the precise P-value for determining considered independently, but also is (D.C. cu. 1978). that a study is conclusively positive is considered as part of the totality of the irrelevant to this proceeding. I will not B. Classification ofCarcinogenicity resolve that issue here, but rather will evidence. An individual study, standing Studies alone, may not raise a serious question resolve it when it is presented in the as to the safety of a substance. When Classifications for carcinogenicity context of an administrative proceeding that study is viewed with other similar studies are simply terms used to reflect in which it is relevant. studies, a trend of a particular effect the conclusions drawn from a study. 2. InconclusiveBut Suggesti~'e ofa may become apparent Where several Studies submitted in this proceeding can PositiveEffect. As discussed above in studies, viewed together, point in the be classified as (1)positive, (2) Section mA3.•Abbott contends that all direction of carcinogenicity, those inconclusive but suggestive of a positive studies that are not positive should be studies. even though inconclusive, are a effect, (3)negative, or (4) deficient. considered as negative and cannot be valid and objective basis for concluding These classifications reflect whether a relied upon to deny approval of a food that a food additive has not been shown study supports the conclusion that the additive petition (Abbott's Remand Brief to be safe. This is particularly true when test substance causes cancer (positive), at 12-18). The Bureau contends that an the inability to demonstrate a suggests that the test substance causes inconclusive study may raise serious statistically significant treatment effect cancer (inconclusive but suggestive of a questions as to the safety of cyclamate in the individual studies is a result of the and thus support the conclusion that the insensitivity of the studies. liThe CourtIn EthfJ Corp. wu re\iewlnBEPA'. additive has not been shown to be safe declsionunder the arbitrary and capriclOUl (Bureau's Remand Reply at 5-6). Courts have consistently upheld standard of the AdmtniJtrative ProcedureAct.5 decisions made by federal agencies U.S,c. 706(2)(A) (1976).AlthouP the cyclamate As also discussed above.I find that a declslonIs lub/ed 10re\iew WIder thelubltantia! study which is inconclusive because of .. It shouldbe emphasizedthat the great ma/orlty evidence standard of 5 U.s.c. 706(2)[f:)(1m). the questions about its statistical or ofsubstances donot cause cancerwhen tested In propositionstated above 11 nll\'erther... aJlPUcable biological significance may nevertheless the types ofanimal studiescontainedIn this record. here because It does not relate 10 the appUCable AttentionIn thereforeproperlypaid to suchstudies standard ofreviewbut rather to the applicationat raise a serious doubt as to the safety of whenevercanceroustumorsare found. sclel\tific prlnclples10admtnlstrati\'e!lIclfindlnB. a food additive and be relied on by the , < 61482 Federal Register I Vol. 45. No. 181 I Tuesday. September 16. 1980 I Notices agency as a basis for denial of a food "power" of a statistical test or false In response to the specific question additive petition. negative.error rate is the probability asked in the above-quoted languag of 3.Negative. A negative study is a (frequency) that the test will detect. at a the Remand Order. the Bureau referred study that supports the conclusion of a specified confidence level. a specific ' to a statistical review in the Temporary reasonable certainty of no harm. As minimum difference between treated Committee Report (G-41 App, V at 19 with positive studies, negative studies and control animals. ifthe difference is 20). That statistical review reports the are attributed various weights present For example. the Plank study minimum detectable difference between depending on the statistical and had only a 50%chance ofdetecting. at cyclamate-treated and control animal biological significance of the study. It is the 95%confidence level. a true for each cyclamate carcfnogenlclty important to bear in mind. however. that difference in tumor incidence of study reviewed by the Temporary in view of the serious consequences.of approximately 33%between the controls Committee (Bureau's Position Paper at 5 mistakenly finding that a negative study and the high dose treated animals. The n. 1). The Bureau also cites the proves safety. a flawed negative study "33%"figure in this example is the Interagency Regulatory Liaison Report. may be entitled to little or no weight minimum detectable difference that this which I have not considered because the whereas a positive study with a similar study is capable of detecting at the 95% admission of that report into evidence flaw may well be entitled to some confidence level. The power of this was properly denied by the ALJ (sec . weight. _ study is 50%. This statement tells us that Section VII. F. below). One issue that reoccurs with respect· even ifa true difference in tumor I find that the power of a study and to a number of studies that Abbott incidence of 33%between cyclamate the minimum detectable difference a .considers negative is the sensitivity o~ a treated and control animals existed in study can detect are important criteria study: Abbott recognizes that although a the Plank study. the study would have for determining what. if any. weight study may not detect any effect. itmay only a 50/50 chance of detecting that should be attributed to a study that fails be entitled to little or no weight ifthe difference at the P<.05 confidence level. to detect a statistically significant effect, size 'of the study is so small that the The minimum detectable difference. This method of analysis provides an study is too insensitive to detect an the power of a study and what' objective means of comparing the effect even ifone is in fact present constitutes a statistically significant relative sensitivity of the cyclamate (Abbott's Remand Brief at 18). This .result are dependent on one another and carcinogenicity studios. In my analysis issue can best be understood by on the number of animals in a study (G of the cyclamate carcinogenicity studies .considering a scientist use of a 120 at 4J. Generally. the larger the of questionable sensitivity (see Section microscope. A scientist may be unable number of animals in a.study, the mere IV.B.2.a. (2)-(4): IV.D.) I have therefore to observe an object with a microscope sensitive the study will be. i.e; the lower reported and considered the findings of because the microscope is not powerful the minimum detectable difference the the Temporary Committee ooncerning study can detect at a specified power enough to sufficiently magnify the object 16 the minimum detectable difference each to make it visible. Similarly, a small and confidence level. ' Abbott contends that "ifno cyclamate carcinogenicity study is study may be too insensitive to detect a capable of detecting. carcinogenic effect. even though one is statistically significant (p<.05) effects are observed in a study then itis. Itshould be noted that. in determining present. In evaluating carcinogenicity the minimum detectable difference studies. statistical methodology is used negative: however. all negatives are not of equal value" (Abbott's Remand Brief between cyclamate-treated animals and to determine the likelihood that a real control animals in the cyclamate effect is present even though the study at18). Abbott does not, however.· articulate what it considers to be the carcinogenicity studies. the Temporary did not detect any effect. In the Remand Committee (lJ assumed that the power Order. Commissioner Kennedy asked criteria for determining whether the sensitivity of a study is adequate. of each study was 50%. (2) assumed that the parties to further explain their statistical significance was P<.05. and positions on this issue: . Abbott states only that "commonly accepted scientific standards for (3) reported the resulting minimum Another issue that needs further ' determining safety are well known and detectable difference for each study. For development by the parties concerns criteria understood" (Abbott's Remand Brief at example. the Temporary Committee for determining proof of safety. This . 18; A-858 at 25).Abbott also lists the reported that the Ikeda study (mscssed determination involves an assessment of the Schmaehl, Kroes; Taylor. Gaunt and below) had only a 50%change of quality of a study which in turn involves two detecting. at the 95%confidence level, a main considerations: the minimum difference -, Carson studies as examples of negative that a study can detect between effects on studies providing proof that cyclamate is true difference in tumor incidence of control animals and effects on treated safe. Although I agree that the Gaimt approximately 13%between the controls animals. and the frequency with which this and Carson studies are negative. I and the high dose treated animals (G-41 difference can be detected. Abbott appears to disagree with the remainder of that App, V at 20).I do not consider the 50% argue that any study not significant at the statement. My reasons are discussed power utilized by the Temporary ".05 confidence levelis negative and should below in Section IV. Committee to be an especially high one. be considered as proof of safe~ regardless of It means that 50%of the time, when the the sensitivity of the test or the frequency '"The Bureau also notes that the power of a test specified minimum detectable difference with the which the study would detect a "depends on how exaggerated the highest dose is actually present it will not be specified difference. studied is compared to the estimate of human consumption"(Bureau'sPositionPaper at 5 n, 1). declared significant at the P<.05 level. (44 FR 47622). Thlsstatement is incorrect.The statistical power of Given the consequences of incorrectly The terms referred to by a study will remain constant even thoughthe dose declaring that a study is negative, I do Commissioner Kennedy are used to studied may vary. ITthe Bureau means to suggest not find a potential false negative error .that a study may have an adequate statistical power describe the sensitivity of a study. The . but nevertheless be inadequate because the highest rate of 50%to be very reassuring. I find• term "minimum difference" refers to the dose studied is too low. I agree. However. the however. that even assuming that a 50% minimum difference between treated Bureauhas not criticized any of the dose levels power is adequate. the minimum and control animals that a study is employ"~d in the cyclamate earcinogenJciiY studied as'being too low. Nor.for that matter; has Abbott detectable difference in the cyclamate capable of detecting at a specified' criticized any of the dose levels studies as being too carcinogenicity studies of questionable confidence level and frequency. The high. " sensitivity is unacceptably high. Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61483
# I recognize that it is impossible to .. that the National Cancer Institute S.Short·termor in \ilro test systems cannot prove a negative to an absolute ("NCI") establish an advlsory committee now bewed to establish carcinogenicity. certainty and I am not asking Abbott to of experts to review the carclnogenlclty However, the results fromsuch systems are weful for determiningthe need for do so. However, I disagree with Abbott evidence concerning cyclamate and appropriate carcinogenbioassay studies. as as to what weight, if any, should be advise the agency as to whether or not well as for enlargingthe mutagenicity attributed to many of the studies that cyclamate is a carcinogen (G-41, App 1). carcinogenicitycorrelativedata base. In this Abbott considers negative (see Sections The National Cancer Institute thereafter regard. the Committee notes that in several IV.B.2.a.(It-(5) and IV.D.).For the established a Temporary Committee for studies eyclamate or cyclohexylamine has reasons discussed below, I find that the Review of Data on Carcinogenicity been found to produce chromosomal damage many of the studies that Abbott of Cyclamate (''Temporary Committee") in human and rodent cells. contends are negative, do not provide a to advise NCI concerning its scientific (G-41 at 48). basis for any valid conclusions as to the review on all available data on the The adnceof the TemporcuJ' safety of cyclamate because of the low carcinogenicity of cyclamate. The Commillee is, of course, not controlling sensitivity of those studies. Indeed, I Temporary Committee consisted of a in this proceeding. The Temporary have found that there are only two number of distinguished scientists, Commillee's conclusions are, however. studies (Gaunt and Carson) submitted including oncologists, pathologists, e..tidence in this proceeding and should by Abbott that are properly classified as medical doctors and doctors of be considered as such. Abbott contends negative (see Section IV.C.). These two veterinary medicine. In addition, four that the Temporary Committee could not negative studies are not. however, working groups were established to have made "a more definitive statement entitled to sufficient weight to meet provide staff support and additional regarding cyclamate's safety" (Abboll's Abbott's burden of proving to a expertise to the Temporary Committee, Remand Ex. at 9). I disagree. A much reasonable certainty that cyclamate These working groups included the NCI more definitive statement could have does not cause cancer nor do they rebut Epidemiology Working Group, the NCI been written, namely that cyclamate has the safety questions raised by other Experimental Design and Toxicology been shown to be safe. The Temporary studies (see Section IV.C). ' Working Group, the NCI Pathology Committe did not make such a finding. 4. Deficient. A study may be deficient Working Group and the NCI Statistics Indeed, as is apparent from paragraphs because of defects in the design or Working Group. 3 and 4 above, the Temporary conduct of the study. The parties do not In February, 1976, the Temporary Committee expressed substantial dispute that where a study contains a Committe submitted its report to the uncertainty about the safety of significant defect it should not be given Director of the National Cancer cyclamate. In addition to the statements any weight The parties also agree that Institute, The Thmporary Committee in paragraphs 3 and 4, the Temporary even where the conduct of a study is not concluded that: Committee stated that defective, that study may be entitled to 1.The present evidence does not establish None of those sludies (referring to the Bryan, no weight because it is too insensitive to the carcinogenicity of cyclamateor its provide any useful information about Oser, Hicks and Friedmanstudies) satisfy the principal metabolite.cyclohexylamine, in Commillee'scriteria for concludingthat the safety of the test substance experimentalanimals. cyclamate is a carcinogen.TheyQ'O, however, (Abbott's Remand Brief at 18). Abbott 2. No conclusionscan be made regarding create a sense o£uncertainty. contends that inadequately sensitive the question of cyclamate's potential studies sheuld be classified as negative, carcinogenicity in humans due to the short G-41 at 46. Moreover, the Experimental although entitled to little or no weight post-exposureobservation time,the Design and Toxicology Working Group insensitivityof epidemiologic studies to (id]. The Bureau contends that a study of the Temporary Committee found that detect relativelysmall changes in cancer "the studies thus far conducted have of inadequate sensitivity or an incidence,and other faclors. otherwise deficient study should be 3.The Committee i. concernedover the been inadequate to assess the classified "inconclusive but implications of the increased incidenceof carcinogenicity of cyclamate in animals". uninformative" (Bureau's Position Paper tumorsin the urinary tract of cyclamate-Jed (G-41, App. V at 55). at 7). I find that there is no substantive animals fromseveral studies, even though The Temporary Committee also difference in these approaches, but only those increases were not statistically described a study designed to resolve a question or nomenclature. I have significant. It is not clear whether this the Committee's uncertainty about the represents a weak carcinogenicresponse or decided to classify such studies as random variation. _ safety of cyclamate. Abbott contends that requests for additional testing result deficient . 4. An additional concern Is the in a "never-satisfied posture" in view of IV. Carcinogenicity: The Evidence carcinogenicresponses obtained in cyclamate-treatedanimals fromstudies the Temporary Committee's statement With the principles discussed in employing unconventionalprocedures or in that U[c]yclamate has pushed the Sections II and ill in mind. I will now which the specificityof the response i. technology of carcinogenicity testing to discuss the evidence submitted in this questionable.The bladder implantationstudy its limit" (G-41 at 47). I disagree. The proceeding. One piece of evidence that done by Bryanet al, was considered to be statute places on Abbott the burden of inappropriatefor assessing carcinogenicity o£ was the subject of much dispute was the a human dietary constituent.Of particular proving that cyclamate is safe. Congress . Report of the Temporary Committee For concern is the Food and DrugResearch has thus decided that where the The Review of Data On Carcinogenicity Laboratories'study (Oser et al.) Inwhich a evidence is uncertain the petition must of Cyclamate (G-41). Because this report slatistically significantincrease in bladder be denied, regardless of whether is cited by both parties as part of their tumorsoccurredin animals treated with a additional testing could resolve that discussion of most of the carcinogenicity mixlureof cyclamate and saccharin.The uncertainty. The fact that the studies. I will discuss it first cocarcinogenicily system used by Hicks et al, "uncertainty [about cyclamate's safety] has yet to be validated as a bioassay for does not appear to be easily resolvable A. The Reviewofthe Temporary " carcinogenicity. Althoughthe dose-dependent Committee ofthe National Cancer increase in lymphosarcomas in cyclamate by currently available bioassay Institute treated mice (Brnntom et al.)was statistically technology" (G-41 at 46) does not lessen significant, there Is the likelihoodthat this Abbott's burden. On March 14, 1975. then reflects a nonspecificresponse in the straln In the case orcyclamate, itis certainly Commissioner A.M.Schmidt requested of miceemployed. possible that further adequate testing, 61484 Federal Rllgisler I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices
such as the study proposed py the 30 female mice of the MIG strain and an lesions visible (without histopathology) Temporary Committee, could resolve the equal number of female controls: and 40 male intreated animals but not in controls .current questions about cyclamate's laboratory-bred mice of the F1 (C3H x Rill) are at the least an indication of a more possible carcinogenicity. Ifsuch testing strain and an equal number of controls. The rapid onset of the effect seen and are study wall conducted for the lifetime of the is done, it may yet be possible for FDA animals. The animals were examined grossly evidence of a greater chance for to conclude that there is a reasonable but special attention was not given to the metastasis (spread of cancer) (id.). The certainty that cyclamate does not cause bladders nor were bladders.examined' Bureau argues that even if cancer. histopathologic~lly. histophathology revealed some tumors B. Inconclusive but Suggestive Studies' (ID at 10). in the control group. those tumors would Raising a Serious Question as to the (2) Study Results: The authors-of the have been smaller and later in 'Possible Carcinogenicity ofCyclamate Rudali study concluded that cyclamate developing. These factors, the Bureau concludes, make the lung tumor findings 1. The Occurrence tJ!LungandLive; is a weak carcinogen (A-412 at 3). The ALJ found that in the first Rudali study, in the Rudali study toxicologically Tumors andLymphosarcomas in Mice. significant, even though no The Rudali, Brantom, Kroes and Hardy an increased incidence and shortened latency was seen for lung tumors in histopathology was performed (id. at 3). studies all involve the direct feeding of Abbott, relying on the testimony of Dr. cyclamate to test animals and suggest XVII/G reated female mice (ID at10). The incidence of these lung tumors was Smuckler, contends that due to the lack that cyclamate is a carcinogen. In the of histopathology the Rudali study Rudali study, one strain of cyclamate statistically significant at P=.OOO3. In .the F1 (C3H x RIll) treated male mice, contributes nothing to the assessment of treated female mice was found to have a the potential carcinogenicity of statistically significant incidence at the an increased incidence of hepatomas was seen (id). The incidence of these cyclamate (Abbott's Remand Reply at 4 P<;.051evel of lung tumors and of total 5). Abbott further contends that the tumors combined. A different strain of liver tumors was significant at P=.07. In addition, the incidence of total tumors possibility of metastasis is purely cyclamate treated male mice in the . speculative (id.). Rudali study was found to have an combined in XVII/G treatqd female The Bureau also takes exception to increased incidence (p=.07) of liver mice and F1 (C3H x RIll) male mice the ALI's finding that lung and liver tumors and a statistically slgnificant were statistically significant at P<;.05. tumors in the Rudali study cannot be incidence at P<.05 of total tumors Most of the liver andlung tumors found combined. The Brantom, Kroes and were multiple (A-412 at 2-3). combined. (Dr. Frankos combined total Hardy studies all resulted inincreased In theInltlal Decislon, the ALJ noted tumors, which included lung an liver levels of lymphosarcomas (a malignant that, "[e]ven though an incidence of tumors, in his analysis of the Rudali tumor) in treated animals. In the tumors was seen, the study is deficient' study and found them to be significant Brantom study, there was a statistically in that not all the animals and all their (R. Tr. at 193-96).) The Bureau contends significant dose response relationship organs were subjected to that combining data on lung and liver (P=.OO8) between cyclamate and histopathologic examination" (ID at tumors is permissible and that the lymphosarcomas for female mice and 11).18The Temporary Committee resulting data are biologically . the total incidence of reticuloendothelial reached the same conclusion (G-41 at significant (Bureau's Remand Ex.at 4). sarcomas (p=.06) was biologically 22). . In response, Abbott cites testimony of significant for female mice. In the Kroes Following the reopened hearing, the Dr. Carlborg who states that the study, the incidence of lymphosarcomas ALJ found that "[t]he lung tumor National Cancer Institute has not for three generations of male mice incidence in the Rudali study tends to adopted the practice-of combining combined was statistically significant indict cyclamate as a carcinogen" but tumors from different biological systems (P=.0036). Finally, in the Hardy study, that due to lack of histopathology, . and that he has confirmed this fact with although the incidence of "possible microscopic tumors present In a Ken Chu of NCI (Abbott's Remand lymphosarcomas was not statistically the control animals could have been . Reply at 5-6). significant at the P<;.051evel, it is missed" and that therefore "the Abbott takes exception to the ALJ's important because the study used the biological significance of the Rudali lung finding of a borderline significant effect same mouse strain as the Brantom data is compromised" (IRD at 20). for liver tumors (Abbott's Remand Ex.at study. Thus, the increased The ALJ further found that: 19) Abbott contends that (1)finding is lymphosarcoma levels in the Hardy In order to accept the overall statistical -limited to onesex andone strain: (2) tho study enhance the credibility of the significance of.the lung andliver tumor effect is not significant at tho P<.051ovel results of the Brantom study. Each of levels. data from different biological systems and therefore a higher standard is being these studies is discussed in detail and different mouse strains must be applied to cyclamate than any other below. combined. The controversy over the propriety food additive: and (3) lack of of such combinations would not allow a.RudaJi, et al. (G-43). (1)Study histopathology compromises this finding labeling the overall data biologically (id at 19-20): Design: 17 The ALJ described the Rudali significant. Thus. the borderline significance study as follows: level for liver.tumors is the only biologically I find that the statistically significant (at the P=.OO31evel) incidence o£Iung Sodium cyclamate was placed inthe slgnlflcant effect. . drinking water of several strains of mice at a tumors in the XVII/G female mice in tho (id·r ' Rudali study is a key finding suggosting concentration of 6 gm/liter. A breakdown of (3)Analysis: The Bureau takes the test animals is as follows: 30 male mice of a possible carcinogenic effect of . the Rill strain and an equal number of male exception to the ALI's finding that the cyclamate. This flnding is sufficient by controls: 20 mice and 20 female mice of the lack of histopathology compromised the itself to raise a senous question about C3H strain and an equal number of controls: Rudali study (bureau's Remand E.'C. at 2 the carcinogenicity of cyclamate. Tho 3). The Bureau contends that large lung finding of increased incidence of liver 17Thlsand other descriptions of Study Designs tumors (significant at the P=.071evel) in are laken essenUalIy without change from the ALfs , ..Hlstopathlologic examination refers 10 the Initial Decision. They are iIicIud.~ here to assist the process by which tissues are dried. sectioned. cyclamate treated F1 (C3H X Rill) malo reader In understanding the analysis of the study stained. placed on slides, and examined undera mice and the statistically significant at results. . microscope. . P<;.05increase in total tumors combined Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61485
in cyclamate treated F1 (C3H x RIll) type of pulmonary tumor found need Dr. Frankos received a Ph.D. in19/7 male mice and x\lI{G female mice are clarification" (A-859 at 9-10). Dr. from the University ofMaryland. S.;hool also important because they reinforce Smuckler, however, docs not even DC Pharmacy, Department of - the concerns about the carcinogenicity suggest why the absence oC l~"mphomils, Phrumacology and Toxicology. where he of cyclamate arising from the key even ifunusual, would negate the had experience in the area of finding oflung tumors in the xvn/G observed significant difference in the experimental toxicology of drugs (R. Tr. female mice (G-140 at 10-11). This total evidence of lung and liver tumors at 79-81). This experience is relevant to tumor finding is appropriately relied on between the treated and control groups. the evaluation of the safety of other as part of the overall basis for As to the second part of Dr. Smuckler-'s chemicals (id. at81). From 1977-79, Dr. - concluding that a serious question has statement, the type of pulmonary tumor Frankos worked as a toxicologistin been raised as to the possible present is irrelevant so long as that FDA's Division of Toxicology, Bureau of carcinogenicity of cyclamate, tumor is malignant. I find that the Foods. In that C«lpacity. Dr. Frankos I agree with Abbott that the Bureau's macroscopic examination DC tumors reviewed a total of approximately 100 argument concerning the possibility of confirmed in part by histopathologic toxicity studies (including _metastasis of the tumors found in the examinations fully supports the carcinogenicity studies) submitted in Rudali study is speculative. I do not, conclusion that the lung and llver support ofcompounds for which however. find persuasive Abbott's tumors found by Rudali were malignant. industry firms sought FDA approval (G argument that the lack of histopathology Accordingly, I reject Dr. Smuclder's 140 at2; R. Tr. at 99-100). Dr. Frankos invalidates the findings of liver and lung criticism of this study. has also participated in the design of tumors in this study. Although Abbott also attacks the credibility and toxicology studies (R. Tr. at81-86; 96 histopathology may have revealed reliability of Dr. Frankos, a Bureau 98). Dr. Frankos demonstrated a detailed tumors in control animals. it is equally witness. Although I agree with Abbott knowledge of the type of studies that the possible that it would also have that Dr. Frankos' opinion regarding the Bureau of Foods receives in support of revealed more tumors in treated possible occurrence ofmetastasis in the food additive petitions (R. Tr. at 88-93: animals. Moreover. lung and liver Rudali study was speculative, I 100-101). Abbott surely cannot be tumors that were found macroscopically emphatically reject Abbott's contentions suggesting that experience gainedby a were examined microscopically (A-412 that "Dr. Frankos' testimony is brought scientist serving in a federal regulatory at 2-3). The Site Visit Committee stated into question in virtually every answer agency is DC no value. that histologic. confirmation of all tumors during his cross-examination": "that Fmally. the ALJ did notmake any is essential (G-41 App. m. Foundation Judge Davidson accorded little weight" finding that Dr. Frankos lacked Curie at 4). Although the Site Visit to Dr. Frankos' testimony and that Dr. credibility and did make a number of Committee stated that the quality of Frankos is "inexperienced" (Abbott's findings that were supported by Dr. slides available was generally poor. the Remand Reply at 4). A careful review of Frankos' testimony: e.g., the ALJ found Site Visit Committee did confirm a the testimony of Dr. Frankos and his that the borderline significantle1:elfor number of the lung and liver tumors curriculum vitae reveals that Dr. liver tumors in the Rudali study is found in the Rudali study from a sample Frankos has substantial experience in biologically significant (IRD at 2O)and of the slides (id.). Thismicroscopic the evaluation ofcarcinogenicity studies that the findings ofIymphosarcomas in confirmation of the tumor findings in the and that the cross-examination of Dr. the Brantom study are biologically Frankos, ifaDytAf.ng. enhanoed his significant (IRD at 22). Accmdingly, I Rudali study supports the validity of the credibility.ll macroscopic examinations of lung and • reject Abbott's criticism ofDr. Franko&" liver tumors found in theRudali study. (ld. at 96-97). and fmd that his testimonyis entitledto Finally, I agree with the Bureau's substantial weight. Abbott further contends that since the contention that the large lung and liver "For UUlpIc, Dr.1'JukoIt IMIiAId lila\: lesions. visible without histopathology, At the Bureallor Foock Ilpcta __bet- or years incidence ofliver tumors found in the F1 found in cyclamate treated animals but heJpiDg to de. the Pl'Otoco1l for ItuWeJ ~t will (C3H x RIll) male mice in the Rudali not in controls are an indication of a be considered adequate for JUbmlJaion In the Cl'clic study is not significant at the P<.05 review that 11 soins to beinillat.d In the Dunnor level, a higher standard is being applied more rapid time of onset of the tumors Foods.This WM one of myprime jobJ then. wrilUlg found in cyclamate treated animals (R. - qualitya_lintfactors for the protocols:also to cyclamate than is applied to other Tr. at 188; see also G-140 at 10;R. Tr. at wriUng up. d"lanllll the protoco1lthat we arc ~lrg food additives. I do not find this 190; A-412 at 3). This factor, by itself, is to require the pl!lilJOIltrltO J'Obmit to nl under argument convincing. \\'hen the cyclicreview. incidence oClivertumors in the treated supportive of a finding of • • • And peoplewould cometo me and 15kme carcinogenicity. Thus. even if how would you desisn thisuperilMllt. And1would mice is compared to the incidence in histopathologic examination of the lungs customdeaisn thlngJ ••• (R. Tr. at llZ~): and controls, the P·\'alue is .w. Thus, there is and livers of the mice in the Rudali Q. Butwby does It require lnnovath-etlunl;lngif a 935 probability that the increased every Itudy COI\Iists of 50rats of ellth lpedesat study revealed an equal incidence of incidence in liver tumors in treated each of fourlavell? animals is a result of cyclamate lung and liver tumors in treated and A. Well.illl DOtthat limple. YihrD) DU design a control mice, the.more rapid time of ltudy you bave to look at-v.ell. bow mach of this treatment rather than a result ofchance. onset of the tumors in the cyclamate am I goingto have to feed In the ltudy to esliiblah Ii I would have more confidence that these level that it going to beusable in the human treated mice would still raise a serious results were not a random occurrence if population?· • • they were significant at the P,.051eveI. question as to the carcinogenicity of When )'OU Ilvaluatethat !Uta )'OU couldget cyclamate. toxlcologicalelf«ta thlltv.'em1t dae to the a higher standard is being applied to Abbott also relies on the testimony of compoundbecause you delisned the Itu;!y cyclamate than is applied to otherfood Dr. Smuckler, who in addition to Improperly. Youhave to d«ign a Itud)' that takes additives. I do not find this argument Into consldtrltion the nutritionalrequirements of comincing. When the incidence ofliver questioning the lack of histopathology, that animal Youhave to colllkles- the paatabilils. stated that "[sjince mice are notorious Youhave to consider the fUJdinsl fromIUlY-.hroaiO Then you have to incorporate the p:lJl:'U studlel or actue Itudies because those rUld:n,~s y,..tll enz:;mal!l: assa}.. that mJ,sbt be needed. tte p."1J;:~ for the appearance of spontaneous indicate to you. hey. there is an effectin the u\-er.l disease. the absence of lymphoma and hhtopathobg.c slu:iles that w.u be I:eecd til zero bad betterlook very IpKiIit4Uy lit the Ih-erIn tbiJ In on thaI organ.So those are the more i;moya~ve the absence of critical analysis of the stuuy. t}llCS of studies that I am ta11dng about, 61486 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices - tumors in the treated mice is compared RIIIJ strain of male mice and the XVII!G the biological significance of those two to the incidence in controls, the P-value . strain of female mice. borderline effects. The finding of an is .07.Thus, there is a 93% probability It is appropriate to use the most .increased incidence of a specific typo of that the increased incidence.ln liver sensitive strain of a species for detecting malignant tumor in ~ specific location tumors in treated animals is a result of a toxic effect (G-140 at 11-12), because (such"as the lung tumors found in XVIII cyclamate treatment rather than a result the induction of cancer in any strain or G female mice) is more defmitive than of chance. I would have more species is a good Indicationthat the findings of generalized increased confidence that these results.were not a chemical will probably cause cancer of malignancies (such as the combined random occurrence if they were some type in huinans (icl). Even though total tumors in the Rudali study), but the significant at the P<.05 level. I do not, a tumor finding may be limited to a generalized fmding is still entitled to however, consider these results to be specific species, strain, sex and organ, some weight. insignificant. I agree with the Bureau that finding cannot be dismissed as One of Abbott's witnesses. Dr. and the ALJ that these results are being irrelevant to humans (icl). Absent Carlborg, a statistician, contends that important (G-140 at 10-11j IRD at 20) data indicating what species or strain is the National Cancer Institute ("NCI") and are at least supportive of the most like man insofar as similarity of has rejected the practice of combining conclusion that cyclamate has not been carcinogenic response to cyclamate is tumors from different biological systems shown to be safe. Moreover, my concerned. I' have to assume in the in its bioassay program (A-857 at 6). Tho consideration of carcinogenic effects. interest of public safety that the only support Dr. Carlborg provided for such as the liver tumors found in the response in the most sensitive species, this statement was an experience he had Rudali study, which are not significant strain and sex is most like that of man in which he combined rumors from at P<.05 level, does not impose a higher (id. at 12). different biological systems in a Iltudy of standard on cyclamate than the agency Moreover, it is not entirely true that toxaphene (R. Tr. at 48). Dr. Carlborg has imposed on other food additives. As the liver,tumors in F1 (C3H x RIIIJ male testified that his analysis of the the discussion in Section m establishes, mice and lung tumors in XVII!G female combined tumors resulted in a finding of effects may have biological significance mice were found in only one sex and . no effect, Le; "the tumor rates in the even though they are not statistically only one strain. The combined·incidence control and all the treated groups were significant at the P<.05 level. Although of total tumors, which consisted exactly the same" (id.). Dr. Carlborg such effects are not entitled to as much primarily of lung and liver tumors, in F1 stated that his practice in the case of weight as effects which are significant (C3H x RIIIJ male mice and XVII!G toxaphene was rejected py NCI (id.). at the P<.05Ievel, they are nevertheless female mice, were statlstlcally The example provided by Dr. Carlborg entitled to some weight especially when significant at the P<.05Ievel. Thus, there is. however, distinguishable from tho procedure employed with the lung and considered together with other is evidence that Rudali found an liver tumor data in the Rudall study, As statistically significant results. increased incidence of liver tumors in Dr. Frankos testified, it Is invalid to two different strains and sexes of mice Abbott also argues that the liver combine all tumors to obliterate an tumors found in the F1 (C3H x RIIIJ male and an increased incidence of lung effect (R. Tr. at 194-95). Thus, it Is not. mice are not significant because they tumors in two different sexes and surprising that NCI rejected Dr. were found in only one strain and one strains of mice. Finally, it should be Carlborg's combination of tumors where" 'sex. Presumably, Abbott would make noted that Rudali did not test F1 (C3H x it resulted in a finding of no effect. . the same contention with respect to the RIIIJ female mice or XVII!G male mice. Even ifNCI does not accept the lung tumors found in females of the "'rhus, it is possible that, if tested, the practice of combining tumors from XVII!G strain of mice in the Rudali male XVII!G mice and the female F1 different organ SUIlS where a study. I do not find these contentions (C3H x RIIIJ mice would have exhibited statistically significant (at the P<.05 convincing. The-slgnlflcance of a tumor the same response as their counterparts. level) effect is found, I fmd that the finding in one strain and sex of a species I further disagree with Abbott's method used to analyze the data from is not reduced where that effect does not contention and the ALI's conclusion that the Rudali study is valid. I recognize _ occur in other strains or sexes of the It is inappropriate to combine total that this method does not provide same species. In order to negate tumor tumors (which consisted primarily of conclusive evidence of cyclamate's findings in a particular strain and-sex of lung and liver tumors) found in the same carcinogenicity. However, it does a species, it is necessary to.conduct strain of mice in the Rudali study, for contribute to the assessment of further studies in the same strain and the purpose of obtaining additional cyclamate's carcinogenicity and raises a sex of the species in which the tumor information about the potential' serious question as to the possible finding was made. Such testing is carcinogenicity of cyclamate. Combining carcinogenicity of cyclamate. necessary because it is not unusual to tumors from different organ sites is ' In sum, I find that'the Rudali study find more of an effect in a particular sex appropriate in order to evaluate suggests, but does not prove, that or a particular strain (R. Tr. at 107-{)8j cyclamate's overall carcinogenic . cyclamate is a carcinogen. G-140 at 11-12). The fact that other potential (R. Tr. at 193-96j see G-118 at b. Brantom, et al,(G-3). (1) Study strains or sexes of mice tested by Rudali 18-19). This approach is particularly Design: This study involved groups. of 30 did not exhibit the same lung and liver' valid where. as here. a statistically male and 30 female mice fed .7,1.75, 3.5 tumor effect does not lessen the significant tumor Increase is seen in one or 7.0% sodium cyclamate. A control ' significance of the liver tumors found in organ (lung) in one strain of mice (XVII! group of 60 mice of each sex was F1 (C3H x RIll) strain of male mice and G) and borderline significant tumor maintained. The study was continued lung tumors found in the xvtutc strain increase is seen in the sameorgan and a for 80 weeks, after which survivors were of female mice. Even within the same second organ (liver) in a second strain sacrificed. species, strains or sexes can vary in of mice (F1 (C3H x RID)) (R•.'fr. at 193 (2) Study Results:In the Initial sensitivity (G-140 at 11j R. Tr. at 107-{)8). 96). The finding of statistical Decision, the ALJ found that "a Thus, to negate the lung and liver tumor SIgnificance for total tumors in the strain statistically significant increase of findings in the Rudali study, further of mice with two borderline effects lymphosarcomas was found in tho testing must be done in ~e F1 (C3H x increases the confidence to be placed on Brantom study" (ill at 31). Following tho . -, . -Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61487 ; reopened hearing, the ALI made the significant at the P=.076Ic\·e), and the in more than one study is a factor that following finding with respect to the incidence of lymphosarcomas and should be considered in determining the Brantom study: reticulum cell sarcomas, for female biological significance of a borderline • • • the Bureauy found a biologically mice, which was significant at the P=.06 significant effect (G-140 at13). I significant effect for cyclamate in the total level, are biologically significant therefore conclude that the incidence of incidences of lymphosarcomas and reticulum (Bureau's Remand Reply at 5). lymphosarcomas and reticulum-cell cell sarcomas in the female treated groups I find that the incidence of sarcomas combined found in the when compared to the controls (Ex. No. G lymphosarcomas and the incidence of Brantom study are biologically 140 at 7).Abbott challenges this data because lymphosarcomas and reticulum-cell significant. the Bonferroni multiplier was not applied. . sarcomas combined are key fmdings It is important to note that the P-vaIue Even ifthis multiplier is used, however, two that suggest that cyclamate is a of.076 cited by the for the linear figures remain of borderline statistical ALI significance [Linear trend for carcinogen. There is a statistically trend for lymphosarcomas is erroneous. lymphosarcomas and the incidence of significant (p=.008) dose response Thisfigurewasarrivedatbyapp~ing lymphosarcomas and reticulum cell sarcomas relationship between cyclamate and the the Bonferroni correction to the P-vaIue combined}. One of these two effects is also incidence of lymphosarcomas in female for the linear trend for lymphosarcomas. challenged for failing to properly use the mice in the Brantom study (G-139 at 6). However, as Abbott's witness. Dr. Armitage test (Ex. No. A-857 at 13)!-Buteven Moreover, even accepting Abbott's Garlborg conceded. the Bonferroni - assuming the validity of this challenge. a statistical analysis of the data, the correction is applied only to indi...idnal borderline statistically significant effect of incidence of lymphosarcomas and comparisons and not to.trend tests and the remaining figure. for the total reticulum cell sarcomas combined is (R. 33}. reticuloendothelial sarcoma rates.exists, dose responses Tr. at Thus. the When considered in conjunction with the significant at the P=.OOleveL I agree Bonferroni multiplier offour was dose related increase in lymphosarcomas for with the ALI that the dose response improperly applied to the linear trend female treated animals, this trend renders the relationship in female mice, when for lymphosarcomas and the correct P Brantom data biologically significant. viewed with the borderline statistically value Is .019.Although Dr. Carlborg significant incidence for all (IRD at 21-22). criticizes this result because it was The Temporary Committee made the reticulonendothelial sarcomas, renders achieved by use of the Armitage test, the Brantom data biologically which he claims is inappropriate for the following finding with respect to the significant. . Brantom study: lymphoearcoma finding. Dr. Carlberg I reject Dr. Carlborg's statement that does not state that the result would be • • "the Committee agrees that the test ''when the multiplier of4 is applied to any different if the method he claims is material did not induce a carcinogenic [the P-value for lymphosarcomas and correct were used. Moreover, although response in the urinary bladders ofthe treated animals. Although the increased reticulum-cell sarcomas combined], the Dr. Carlberg identified all linear trend incidence oflymphosarcomas in the P-value is .060 (4X.015). and any tests which he thought were cyclamate-fedfemale mice requires close significance vanishes." (A-857 at 13). inappropriate (A-857 at13). he did not evaluation, the nonspecific nature of this Even assuming that the use of this state that the Armitage test was response makes its significance questionable Bonferroni multiplier is valid, there Is no inappropriate for analyzing the linear with respect to establishing carcinogenicity. basis in science for the proposition that trend for lymphosarcomas and (G-41 at 16). the potential carcinogenic effect reticulum-cell sarcomas combined (icl.). (3)Analysis: In its exceptions. Abbott "vanishes" simply because the P-value That trend test was statistically contends that the two findings which the is greater than .05.There is no significant at P=.045. Finally, Dr. Gaylor ALI found to be at ''borderline statistical qualitative difference between a P-value found that the dose response significance" (if the statistical of .05 and .06.The difference is merely relationship between cyclamate and corrections insisted on by Abbott are quantitative. To suggest thal the lymphosarcomas in the Brantom study applied) are negative based on relavitely small quantitative difference was significant at P=.OO8(G-139 at 6) established Bureau criteria (Abbott's between a P-value of .05 and .06renders and his statistical methodology was not Remand Ex. at 26). the resulting data meaningless is to challenged. The borderline findings to which ignore the scientific realities of the Abbott also contends here. as it does Abbott refers are the increased situation.~ with respect to the Kroes study (in incidence of combined lymphosarcomas I find that the strong dose-response which a statistically significant and reticulum cell sarcomas (p=.06) and relationship between cyclamate and the incidence oflymphosarcomas was the linear trend for lymphosarcomas incidence of lymphosarcomas (p=.008) found), that the incidence of (p=.076) (linear trend is a statistical test and the linear trend for lymphosarcomas lymphosarcomas and the dose response used to test for presence of a dose and reticulum cell sarcomas combined relationship are artifacts. Le••chance response relationship). Abbott further (P=.045) support the conclusion that the occurrences. Abbott contends that this contends that the lymphosarcoma and incidence of lymphosarcomas and result is due to the "infinite number of reticulum cell sarcoma finding in the reticulum cell sarcomas combined are comparisons [that) canbe made" Brantom study "was a chance biologically significant. In addition. the (Abbott's Remand ~ at 26).Abbott occurrence such as is bound to arise in findings of lymphosarcomas in the Kroes also relies on the fact that the chance of such a vast amount of data" (id. at 27). and Hardy studies also support the an arithmetic decrease in The Bureau contends that the key conclusion that the lymphosarcomas lymphosarcomas inmale mice in the finding in the Brantom data is the dose and reticulum cell sarcomas in the Brantom study is 1 in 120 (exactly the response relationship between Brantom study are biologically opposite of the increase found in female cyclamate and lymphosarcomas for significant (G-139 at 9-10: see G-140 at mice) and a statistical analysis ofUver female mice which was statistically 7-8). The occurrence of the same fmding tumors in the Brantom study indicates significant at the P=.OO8level (Bureau's that cyclamate is a carcinogen in Remand Reply ars; G-139 at 6). The ..I have alsumed for the sake of IfI11II1t'nl, wilbout deciding upon ltllntrinslc merits. that the females and an "anticarcinogen" in Bureau also argues that the linear trend Bonferronicorrec:Uon lhouJd be URd in anaJ)'Zinc males (id.). Abbott claims that there is test for lymphosarcomas, which was data such olbat in lbe Branlom Itudy. no known scientific rationale to support 61488 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices
-thevalidity of these inconsistent valid negative results (R. Tr. at 186-87). death, thus preventing meaningful conclusions (id.). Absent such evj.dence, mere speculation histopathological examlnatlon.] I do not flnd Abbott's arguments is insufficient to support a conclusion Following the reopened hearing, tho persuasive. First, there is a scientific: that the findings of lymphosarcomaa and ALI found that "[a] statistically explanation for what Abbott has reticulum cell-sarcomas combined in the significant effect for lymphosarcomas characterized as an "anticarclnogenic" Brantom study are artifacts. exists in the Kroes study if all threo effect~An apparent decrease in tumors I recognize that my conclusion with . treated male generations are compared with increase in dose may be a result of respect to the Brantom.study is contrary with the sum of their control c:ompeting risks of deaths from other to the finding of the authors of the study counterparts" (IRD at 22). The ALJ diseases which obscure the presence of and the Temporary Committee. The furtherfound that "[o]nly if the worst cancer at high doses (R. Tr. at 53). Thus, authors of the study concluded that "the case against cyclamate is assumed. what Abbott claims is cyclamate's incidence oflymphoma was not affected however, does the data withstand "anitcarcinogenic" effect on , by the feeding of cyclamate" (G-3 at Abbott's criticism [that combining the lymphosarcomas and liver tumors may 744). The Temporary Committee found three generations is inappropriate]" (Id.). not be an artifact but may be que to that the significance of the The Temporary Committee found the mortality from other causes (see, e.g. G lymphosarcomas was questionable of study :" * • to have been well designed 41, App. VII,British Industrial Biological. the nonspecific nature of the response and conducted, although its slgnlfloanca Research, Association at 1). (G-41 at 16). . was reduced somewhat as a result of a Even if there were an The conclusions of the aut1i.ors of a subst-antial number of mice lost from "anticarcinogenic" effect in male mice in study that the test results are negative is autolysis ••• [N]one of the test the Brantom study, it would not negate not dispositive (R. Tr. at157). That materials displayed carcinogenicity." the biological significance of the .; conclusion can be rebutted by other (G-41 at 26.) lymphosarcoma and reticulum-cell evidence, for example, a statistical ,(3)Analysis: In its exceptions to the sarcoma findings in the female mice (R. analysis showing some positive results Initial Decision, Abbott contended that Tr. at 182). This is particularly true in that need further investigation, or the signfficance of the Kroes study was view of the occurrence of evidence of a defect Inthe execution of reduced by autolysis, but that the study lymphosarcomas in the Kroes and the study. In the case of the Brantom is not insignificant as a negative study Hardy studies (discussed belo..w). The study, two statistical analyses (Abbott's Exceptions at 29-30). With occurrence of lymphosarcomas in the (nonparametric dose-response and respect to the ALI's fmdings after the Hardy and Kroes study adds credence linear trend) show a stajlstlcally reopened hearing, Abbottconcedes that to the lymphosarcoma finding in the significant effect and an analysis of the lymph system sarcomas in the three Brantom study (G-139 at 6-7; G-140 at Iymphosaroomas.shows biologically combined generations of the male mice 9-10) and tends to negate Abbott's . significant effect. This evidence rebuts in the Kroes study are.statistically argument that the Brantom fmdings are the conclusion of the authors of the artifacts. ' significant at the P " .05 level (Abbott's study and th~ Temporary Committee. It is hard to understand how Abbott Remand Ex. at 27). Abbott contends, and, as discussed above, has not been however, that (1) it is inappropriate to can argue that the reticuloendothelial adequately refuted by Abbott. As to the sarcoma findings in the Brantom study combine these generations because this Temporary Committee's finding that are artifacts resulting from the infinite method has not been emloyed lymphosarcomas were not site specific, I number of possible comparisons in view elsewhere: (2) the effect is sex speciflc of Abbott's application of the Bonferroni agree with Dr. Samuel Epstein, a Bureau for males. but a sex specific effect is not correction. The purposeof the .. witness, who stated that ". • • ,the. confirmed by other studies: (3) the high Bonferroni correction is to adjust for the comments of the [Temporary spontaneous incidence of increased false positive error rate that. Committee] Report that lymphosarcomas easily explains thla can result from multiple comparisons. lymphosarcomas are inconsequential fmding; and (4) the effect is an artifact As the above discussion establishes, because they are 'nonspecific tumors' • because the treated males in another however, even applying the Bonferroni appears'incomprehensible. A study, the Brantom study, experienced correction where Dr. Carlbourg contends lymphosarcoma is a malignant fewer tumors than their controls it should be applied, the effect on the tumor • • ." (G-121 at 6; see G-118 at (Abbott's Remand Ex.at 27-29). reticuloendothelial system is significant 19). .. The Bureau's reply is that (1) Dr. at P=.06 and the dose response c. Kroesvet al. (G-76; A-734). (1) Frankos' testmony on the relationship is significant at P=.008. Study Design: The ALI described the appropriateness of combining Abbott cannot have it both ways. If Kroes study as follows: generations is uncontradicted; (2) tho Abbott wants to correct for multiple This study employed SPF-derived swiss alleged high spontaneous incidence of comparisons. it cannot complain that the mice in groups of 50 animals of each sex. The lymphosarcomas in other studies is resulting figures are nevertheless invalid groups were fed 2 or 5%sodium cyclamate, 2 irrelevant because there is no testimony because of the multiple comparisons ' or 5%cyclamate-saccharin in a 10:1 mixture, that the control incidence of that have been employed. or 0.2 or 0.5%saccharin or 0.5%CHA. A • lymphosarcomas in the Kroes study is control group of equal size was also unusually low; and (3) the finding of The fact that one or more artifacts is maintained. likely to occur in a study such as the :, lymphosarcomas in the Brantom and Brantom study does not prove that a (ill at 10.) Hardy study negate the possibility that particular effect, such as, the (2) Study Results: In the Initial the Kroes finding is an artifact (Bureau's lymphosarcomas, is an artifact. I cannot Decision, the ALI found that "[b]oth Remand Reply at 6-7). The Bureau also disregard a potential carcinogenic effect parties agree that Ute study is negative, contends that autolysis limited based on such a speculative argument. but the Bureau contends that its substantially the detectability of effects .In order to rebut such a finding, it is sensitivity is severely reduced because in the Kroes study, thus limiting the necessary to adequately study the same of the large number of animals lost to sensitivity of the study (Bureau's Brief at sex/strain/species under the same autolysis" (ill at 10). (Autolysis is a 1~; G-121 at 9j G-126 at 12; G-113 at 7: experimental conditions and obtain decay of tissue that begins shortly after G-112 at 15). Federal Register I Vol. 45, No. 181 I Tuesday. September 16, 1980 I Notices 61489
The Bureau also takes exception to Abbott contends, however, that Dr. Remand Ex. at 28). The only evidence of the ALI's criticism of combining Frankos' testimony concerning the use the spontaneous incidence of generations in the Kroes study. The of this method in other studies is lymphosarcomas of the strain ofmice in Bureau contends that the uncontradicted equivocal and should be given no weight the Kroes study is the Temporary testimony establishes that comblning (Abbott Remand Ex. at 21-.28; Abbott Committee Report. (Abbott also cites an the data from generations is appropriate Remand Reply at 2-3). Dr.Frankos exhibit submitted by the Bureau which (Bureau Remand Ex. at 2; R. Tr. at 159 testified that the combining of reports the spontaneous incidence of 50, 164-65). The testimony cited by the generations was employed as a method leukemia-lymphomas as being between Bureau is that of Dr. Frankos who of analyzing data on the possible 1.6 and 6.~ (G-141 at 962). However, testified that he approves of the carcinogenicity ofxylitol (R. Tr. at 168). this report does not involve the same combination of generations because it Abbott contends that just prior to giving strain ofmice as that used in the Kroes increases the sensitivity of the study this testimony Dr. Frankos was study.) The Temporary Committee and is very analogous to the human uncertain about his answer (Abbott's report states that the spontaneous situation of many generations being Remand Ex. at 28). However, Dr. diseases for the strain ofmice used in exposed to a compound (R. Tr. at 164 Frankos' second answer is emphatic and the Kroes study "includes a 5-10%_ 65). I find it has probative value. incidence ofleukemia (primarily I find that the data generated from the Moreover. Dr. Frankos also testified. lymphocytic)" (G-41, App, Ill, National three generations of mice fed cyclamate in response to a question about whether Institute ofPubllc Health, at 2). in the Kroes study were properly the FDA permits reviewers to combine However. the Kroes study reported combined and analyzed and that the generations for review of a leukemias separately from statistically significant (p = .(036) multigeneration study, that "[ojur lymphosarcomas. Therefore, it is unclear lymphosarcoma finding is a key finding statisticians have done it * * *. We whether the spontaneous incidence of that suggests that cyclamate is a have your statistician. Dr. Carlborg and lymphosarcomas in the strain ofmice in carcinogen. Moreover, the finding of Dr. Gaylor and other statisticians, they the Kroes study is in fact 5-10%. The lymphosarcomas and reticulum cell all have done that" (R. Tr. at 160). Thus, three separate generations ofmice in the sarcomas in the Brantom study I find tht Dr. Frankos' testimony, when Kroes study had a zero incidence of reinforces the concerns about the read in its entirety, is credible and lymphosarcomas (R. Tr. at100). This carcinogenicity of cyclamate arising supports the conclusion that the would indicate that the historical from the lymphosarcomas found in the combining of generations in the Kroes incidence oflymphosarcomas in this Kroes study (G-140 at 9-10), study was appropriate. strain of mice is low (id). Even ifthe I reject Abbott's argument that the I also reject Abbott's argument that spontaneous incidence of combination of generations in the Kroes the lymphosarcoma finding in the Kroes lymphosarcomas is 5-10%, there is no study is inappropriate. Dr. Carlborg, study is not biologically significant. testimony that the incidence of who is Abbott's witness and who raised Abbott contends: (1) it is only sex lymphosarcomas in the control mice in every conceivable criticism of the specific in males (not in females and not the Kroes study was unusually low. statistical analyses contained in the in males and females combined). and (2) Thus, the evidence does not establlsh Remand Order, did not criticize the this sex specificity of lymph system that the spontaneous incidence{If combination of generations (A-857). sarcomas is not confirmed by other lymphosarcomas in the strain ofmice Indeed, Dr. Carlborg performed his own studies. Abbott's argument misallocates used in the Kroes study is 5-10% or that statistical analyses of the data utilizing the burden of proof. The burden is not the incidence oflymphosarcomas in the all of the adjustments and types of tests on the Bureau to submit an additional control mice is unusually low. he deemed appropriate, and concluded study conflI'lIling the finding in the Kroes Finally. even assuming that the that when the three generations were study, but rather the burden is on incidence oflymphosercomas in the combined the evidence of Abbott to produce negative results in Kroes study control mice was unusually lymphosarcomas for control vs. male the same sex, species and strain ofmice low, the results of the study were mice treated with 5%cyclamate was as in the Kroes study. The absence of nevertheless statistically significant. significant at the P=.031 level, that increased lymphosarcomas in female Moreover. there was a dose response lymphosarcomas for control vs, all mice in the Kroes study may have been relationship between cyclamate and the cyclamate treated male mice was due to the fact that the survival of the incidence of lymphosarcomas (p=.030) significant at the P=.017 level and the females was significantly less than the (A-857, Exhibit 2 at "Linear trend" for linear trend for male mice was survival of the males (see G-41, App. males). If the tumor difference between significant at the P=.036 level (A-857, VII. National Institute of Publlc Health. cyclamate-treated and control animals Exhibit 2 at Unes 17-18). Although Dr. Netherlands at 1). Moreover, the fact were due to the spontaneous occurrence Carlborg dismisses these statistically that a cancer is found only in a specific Qftumors in treated animals, the effect significant results as "artifacts" he does sex and a specific strain does not mean seen would not be expected to have a not dispute the validity of combining that it can simply be dismissed as being dose response relationship. I cannot generations. Indeed, no Abbott witness irrelevant to humans (G-140 at 11). This conclude that such results are disputes the validity of this method, In issue is discussed in detail in my biologically insignificant absent view of the lack of evidence to the discussion of the Rudall study. For the sufficient additional data in the same contrary, the combining of generations reasons given there, I reject Abbott's strain and sex of mice showing negative by Abbott's own witness, that witness's argument that the lymphosarcomas in results, Abbott here relies on the conclusion that the results were the Kroes study are not biologically testimony ofDr. Carlborg who allegedly statistically significant and thus the significant. found other effects that were artifacts method implicitly valid, and Dr. Finally, I reject Abbott's argument (Abbott's Remand Bx, at 29). This issue Frankos' testimony and the testimony of that the historical spontaneous is discussed in detail in my' discussion of Dr. Gaylor (G-139 at 7) acknowledging incidence of the particular type of tumor the Brantom study. For the reasons the validity of this method, I conclude it in the animal strain in the Kroes study given there, I reject Abbott's argument is a valid method. easily explains the finding (Abbott's that the lymphosarccmas found in the 61490 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices
Kroes study are artifacts and therefore multiplier of 3 to the P-value reported for statement that evidence of cyclamate's ~.'·not biologically significant. lymphosarcomas in the Hardy study, the causing damage to the lymph system . The Bureau also contends that resulting P-value'is .384. Abbott further would be relevant only if cyclamates autolysis limited substantially the notes that there was no dose response were a cancer promoter (id.), Abbott did . detectability of effects in the Kroes relationship exhibited in the Hardy not reply to this exception. study (Bureau's Brief at18). 1agree. study (Abbott's Remand Ex. at 29-30). Itis unclear what the ALI was Autolysis is a decaying of tissue that The Bureau agrees with the ALI that, referring to in the last sentenco pf tho begins shortly after death and that taken by itself, the Hardy study is not above-quoted statement. Several makes examination of tissue more positive. The Bureau argues, however, • matters are, however, clear, difficult. Abbott contends that autolysis that because an increased incidence of Lymphosarcomas are malignant tumors reduced the significance of the negative lymphosarcomas were found in the (G-121 at 6). This evidence is results in the Kroes study, but argues same sex and in the same strain as in • uncontradicted and 1do not believo that that the study is not insignificant as a the Brantom study, the increased the fact that lymphosarcomas are a form negative study. It should flrst be noted incidence is biologically significant of cancer can be seriously disputed that, with respect to lymphosarcomas. 1_ (Bureau's Remand Reply at 6-7; G-139 at (G-140 at 8). have found the results of the Kroes 7). A separate issue, and perhaps the study to be suggestive of 1find that the results of the Hardy issue that caused confusion for the ALI,' carcinogenicity, not-negative. To the study do add to the weight to he given concerns the role played by the lymph extent that the Kroes study did not . the finding oflymphosarcomas in the __ system in immunological defense. An reveal a significant difference between Brantom study which employed the effect on the lymph system could reduce cyclamate-treated and control animals same strain of mice as the Hardy study. an animal's immunological defenses to in the incidence of tumors other than Two factors support this conclusion. an infectious disease causing the animal lymphosarcomas, 1agree with the Brantom used dose level of cyclamate of to die from that disease or to be sick for . Bureau that the autolysis in the Kroes 0, .7, 1.75. 3.5, and 7.7% of the diet. longer periods of time than it might study substantially reduced the Hardy used dose levels of ordinarily (G-140 at 9). This adverse detectability of effects in that study and .cyclohexylamine that were considerably health effect is, however different from thus reduces the sensitivity of the study lower (0••03••10, and .30%) than the cancer, and is not being relied on to (G-121 at 9; G-126 at12; G-113 at 7; G levels of cyclamate used in the Brantom support my findlng that cyclamate has 112 at 15). study. Allowing for the differences in not been shown to be safe, from a d. Hardy, et al. [A-690}. (1) Study dose levels and metabolism of carcinogenicity standpoint, (1 note, , I Design: The ALI described this study as cyclamate to cyclohexylamine, the however, that one study in the record in , follows: responses in the female strain of mice this proceeding examined the effect of calcium cyclamate on the humoral This study employed48 male and 50 used in both the Hardy and Brantom females ASH/CSI (SPF)strain mice. The studies are consistent. immune response ofrabbits and found mice were fed CHA-HCL at concentrations of 1recognize that when the Bonferroni that "cyclamate given to rabbits for 150 300,1,000. or 3,000ppm. A control group of 48 multiplier is applied to the P-value for days increased the period required for males and 50 females was maintained. The' lymphosarcomas in the Hardy study that the immune system to respond to study was conducted for 80 weeks, after the resulting P-value is .384. 1 do not, stimulation by BSA [bovine serum which the survivors were sacrificed. however, find that the results of'the . albumin]" (G-54 at 53). Thus, there is (ill at 16). Hardy study are statistically significant. evidence in the record to support the Thus, the precise P-value selected is not theory that cyclamate may also have an (2)StudyResults:The ALJfound that: This adverse effect on the immune system. Increase • • • was not statistically that important. The important aspect of significant. The Hardy data is important the.Hardy study is that it is biologically 2. The Occurrence ofBladder Tumors because the treated female group reflecting significant in that it supports the flnding in Direct Feeding Studies in Rats. The the increased lymphosarcoma levels was the oflymphosarcomas in the same-mouse occurrence of bladder tumors in a same mouse strain which showed an effect in strain in the Brantom study. number of strains of cyclamate-treated the Brantom study (HSH-eSl) [sic] mice.) e. The Significance of rats in a number of cyclamate direct Although this data enhances the Brantom Lymphosarcomas. The made the feeding studies raised a serious queslion data's credibility. taken alone, it is too ALI following statement concerning the about the safety of cyclamate. Bladder tenuous to warrant declaring cyclamate a tumors in these strains ofrats are rare. carcinogen. significance of lymphosarcomas generally: Their occurrence, even in small numbers (IRD at 22).21 that are not statistically significant at (3) Analysis: Abbott contends that, Evidence was also submitted regarding the P<;.05 withineach study. is biologically applying the Bonferroni inequality potential effects lymphosarcomas'have upon significant. , different body organs and systems. Both , The method employed by the Bureau (Cyclohexyla~ne parties agreed tht because the lymph system , 2' CHA-HCL hydrochloride] is is crucial to an organism's immunological to evaluate the possible carcinogenicity a metabolite of cyclamate. When humans ingest defenses, any assault upon its smooth of cyclamate in these studies was to cyclamate, enzymes in the body may transform (metabolize] some of the cyclamate to functioning threatens that organ's viability; combine a number of studies involving u cyclohexylamine (G-41 at 36).Thus, exposure of a ' However, the parties did not agree that specific strain ofrats and compare the . human to cyclamate may result in exposure to cyclamate was a carcinogen. Onlyif " occurrence of bladder tumors in the cyclohexylamIne also. Several forms of the _ cyclamate was a cancer promoter would cyclamate-treated rats to the cyclamate metabolite. cyclohexylamine. were used these factors be relevant to its safety. in studies that comprise the record of this background rate for that type of tumor proceeding. These forms are cyclohexylamine (IRD at 22). obtained from historical controls. This (CHA], cycloIiexyTamlnehydrochloride (CHA-HCL] The Bureau agrees with the first three method revealed that the difference in and cyclohexylamine sulfate (CHSJ.In the stomach sentences of the above-quoted . tumor incidence between cyclamate all three of these compounds wiII be present in the statement, but takes exception to the same form. For this reason, all three forms of treated animals and historical controls cyclohexylamlne are considered to be biologically final sentence (Bureau's Remand Ex. 5). is statistically significant at P<; .05. As equivalent. and studies using them are relevant in The Bureau contends-that there is no the subsequent discussion establishes, thls proceeding. evidence of record to support the this method.is a valid and scientifically Federal Register I Vol. 45. No. 181 I Tuesday, September 16, 1980 I Notices 61491
acceptable means of evaluating the the Temporary Committee (G-41 at 20 found in cyclamate treated Sprague possible carcinogenicity of a test 21; 25).Moreover. the occurrence of Dawley and Wistar rats and the substance and raises a serious question these tumors in cyclamate treated background rate for such tumors based as to the carcinogenicity of cyclamate. animals is consistent with a small on historical data. a. The Occurrence ofBladder Tumors treatment effect and the occurrence of I recognize that the validity of in Sprague-Dawley and Wistar Rats: these tumors in control animals Is combining the results of different studies The Scbmaehl, Hornberger. Taylor. consistent with the incidence of these and comparing it to historical controls Ikeda. and Hicks (direct feeding) studies tumors in historical controls. The fact can be questioned on the ground that the involved the direct feeding of cyclamate that a similar effect is present in two studies being combined were conducted to Sprague-Dawley or Wistar rats to separate strains ofrats adds credence to in a different manner (G-120 at 11-12). I determine whether cyclamate is a the conclusion that these effects are Iind, however, that the method of carcinogen. (All of these studies are important (Tr. at 613-14).These findings. combining these studies used by the discussed in detail below.) Although by themselves. are biologically Bureau was appropriate for two reasons. cyclamate treated animals in most of significant. First, only studies involving the same these studies did develop tumors. when The statistigal method employed by species and strain of rats were comparing treated and control animals the Bureau confirms that these findings combined (G-120 at 12). This eliminates within Each study, the tumor incidences are biologically significant. It provides the possibility that a strain difference in were not statistically significant at the an objective means of evaluating the the sensltlvlty of these animals to P<:.051evel. The Bureau contends. significance of these rare tumors. The cyclamate would complicate the however. that the three bladder tumors results of the application of this method analysis. Second. the tumor findings in found in cyclamate-treated Sprague to the Wistar and Sprague-Dawley rat the studies that were combined are not Dawley rats in the Hornberger. . strain studies, when viewed together inconsistent (see Tr. at 628).The Schmaehl, and Taylor studies combined with the results of one of the Friedman incidence of bladder tumors in the is statistically significant at the P=.02 studies (discussed below). which control animals in the combined level when compared to the involved Osborne-Mendel rats, led one Sprague-Dawley and Wistar rat studies spontaneous rate of bladder carcinomas Bureau witness, Dr. Charles Brown, who were consistent with each other and in Sprague-Dawley rats (approximately was the head of the statistics working with the incidence ofbladder tumors .23%) based on historical data (G-120 at group for the NCr-Temporary found in control animalsbased on 10; Tr. at 601-604).Moreover. the Bureau Committee. to conclude that "Ior rats. historical data. Thus. the combination of notes that the one bladder tumor found the evidence of positive carcinogenicity the data from these studies is valid. in the control animals in these three is not overwhelming. but it is suggestive It should be emphasized that these studies is not inconsistent 22 with the that they are sensitive to carcinogenic fmdings of bladder tumors in rats do not low background rate based on historical insult by cyclamate" (G-12O at 16). I conclusively establish that cyclamate is data (id). The Bureau further contends agree with Dr. Brown's conclusion. a carcinogen. Moreover, these findings that the three bladder tumors found in do not provide the same degree of In cyclamate-treated Wistar rats in the its exceptions. Abbott does not confidence that one would have ifthe Ikeda and Hicks direct feeding studies contest the propriety ofcombining then results were statistically significant combined is statistically significant at studies. but contends that the results when compared to controls in each the P = .002level when compared to the 'within each study were not statistically study. These findings do. however, raise background rate for bladder carcinomas significant at the P".05level and that a valid and serious question as to (approximately .116%) developed from there was no dose response relationship cyclamate's safety. Itis therefore historical data of the National Cancer (Abbott's Exceptions at 27). Even necessary that cyclamate be tested Institute for all species ofrats combined thought Abbott is correct in its further to resolve this issue. In reaching (G-120 at 10-11). Moreover. the absence characterization of the individual this conclusion, I am not requiring that of any bladder tumors in the control studies. I do not believe this argument Abbott prove a negative. which is. of groups in these two studies is consistent affects the overall significance of the course. impossible. I am. however, with the low background rate based on bladder tumors found in these studies as holding that Abbott cannot escape the historical data [id.]. a group. The sensitivity of most of these force of these studies unless it submits I agree with the Bureau's analyses of studies is low (see G-41. App. V at 19 additional evidence in the form of these data. Although a comparison to 20i see discussion below). Low sufficiently sensitive studies that historical controls would not ordinarily sensitivity is important because. if demonstrate to a reasonable certainty be accepted as a basis for contradicting cyclamate is a weak carcinogen, it that cyclamate does not cause bladder the results of a comparison to would not be expected to produce tumors in rats. concurrent controls within a study, tumors significant at the P<.05Ievel or A detailed discussion of the study where, as here. the individual studies to exhibit a dose response relationship design, study results and my analysis of are of low sensitivity and the tumor in in such small studies (see G-120 at 7-8). the Hicks (direct feeding), Ikeda, Taylor. question has a very low background Indeed. even in the most sensitive of Hornberger, Schmaehl, and Plank rate. such a comparison has validity. these studies. the Schmaehl study, there studies follows. Abbott contends that, What is significant about these studies was a reasonable chance that the study when ...iewed individually, the Hicks is that in a number of studies involving would fail to detect a true difference in (direct feedlngk Ikeda, Taylor. different strains of rats we are seeing tumor incidence of 4~ between control Homberger, Scbmaehl and Plank studies the occurrence of the same rare tumor in animals and those treated at the 5;;; arc negative. As the discussion below treated animals and fewer in controls feeding level (G-120 at6-7). establishes, however, Abbott's (G-121 at 8; see G-12O at 16; G-139 at 6). Accordingly, the lack of a statistically contention is without merit. The importance of the occurrence of significant effect in each of these studies (1) Hicks. et 01. (direct feeding study) such tumors in rats was recognized by when considered alone does not rebut (G-2. A-832). (a) Study Design: This the question about cyclamate's safety study was conducted in conjunction "'The incidence of 1 bladder lumor in 225 total raised by the comparison between the with the Hicks MNU study (discussed control animals is approximately .44%. combined incidence of bladder tumors separately below). The study involved 61492 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices
WistlJr rats fed cyclamate alone. The . 20 days. 3%for days 21-60, 4%for days in cyclamate treated animals. The study was continued for 24 months, after 61-150.5% after 150 days and 6% at one Temporary Committee reported that the which time the animals were sacrificed. year. The study was continued for 28 study was "* * * particularly good In (b) Study Results: Of the 84 surviving months. At the time of the Temporary that animals were exposed in utero and rats fed cyclamate alone, five males , Committee Report. only 40%of the continued on treatment for their were found to have tumors (3 bladder. 2 microscopic examinations had been lifetimes" and that "the test material did kidney). No control animals were found performed.' . not display carcinogenicity" (G-41 at to have tumors. The Pathology Working (b) StudyResults: The Temporary '21). The ALI found that the Taylor study Group of the Temporary Committee Committee found that "none of the test "employed an unacceptably small confirmed three malignant bladder and materials induced tumors of the urinary number of animals pergroup" (IDat 31). two malignant kidney tumors (G-41 at bladder in any of the treated animals" (c)Analysis: Abbott takes exception 25).Although the incidence of these (G-41 at 26).The ALI stated that "[n]o to the ALI's flnding on sfudy size tumors was not statistically significant bladder tumors were observed in the (Abbott's EXceptions at 29).Abbott, at P<.05 (P=.Z). the Temporary animals so far examined. However, relying on the Temporary Committee Committee found that "their occurrence testicular degeneration and urinary Report. contends that the Taylor study Is in a low incidence must be evaluated calculi were observed in treated animals negative (Abbott's Brief at 18, 25).Tho with respect to the reported absence of and appeared to be treatment related" Bureau contends that "the relatively . these tumors in matched and historical (ill at 13-14). _ small number of animals examined control animals" (id). Dr. Hicks stated . (c)Analysis: Abbott relies on the microscopically (for the bladder 49 that the background bladder and kidney report of the Temporary Committee and controls and 53 treated) reduced the tumor rate for these rats in her lab was contends that the study is negative sensitivity of the study" (Bureau's Brief zero (G-114 at 21). , (Abbott's Brief at 24).The Bureau' at 15-16; G-41, App, VII, Taylor and (c)Analysie: Abbott contends that the concedes that no bladder tumors were Friedman 1974 at 2). found in the animals in,the Ikeda study,. fi d th hIt d h I results of this study are not statistically but notes that "histopathology had not I 10 at t e Tay or s u y as on y a significant at the P<.05 level and that been performed on other animal organs 50% chance of detecting, at the 95% therefore the study is negative. I - confidence level. a true difference in disagree. The total.tumor incidence in at the time of the report's publication" tumor incidence of approximately 9% this study is significant at the P=.Z level [Bureau's Brief at 18). The Bureau between the controls and the high dose contends that the Ikeda study is (5%) treated animals (G-41, App, V at (G-114 at 21).There is thus an 80% therefore inconclusive. probability (p=.2) that the results, of the find th th hi th I . th 19). Moreover. the presence of one Hicks direct feeding study are due to I at e lstopa 0 ogy 10 e bladder tumor in a control animal makes Ikeda study is insufficient for classifying the detection of a positive effect more cyclamate instead of a 95% probability this study as negative, particularly in necessary for statistical significance at light of the evidence of lymphosarcomas difficult because the difference between the P<;.05 level. Obviously. I would be. in the Brantom, Hardy and Kroes studies the number of animals with tumors In more certain of the importance of these and the evidence oflung and liver the treated and control groups needs to results if the incidence of bladder tumors in the Rudali study. These be greater in order for that difference to tumors were significapt at the P<.05 studies support the conclusion that be statistically significant [G-112 at 15: level. I do not, however, consider these tumors at sites other than the bladder G-113 at 8; see G-41 App, VII,Taylor results to be negative, particularly in may have been present in the Ikeda and Friedman 1974at 2).This study view of the biological factors present in study, but were not detected. since only therefore is unlikely to detect a small the study. The significance of these _ . bladder histopathology was done. treatment effect. This lock of sensltlvlty results is enhanced by the fact that Dr. Moreover, the Ikeda study has only a is especially important in view of tho Hicks testified that she had never seen 50% chance of detecting, at the 95% findings of bladder tumors in Sprague- such tumors In-untreated animals in her confidence level, a true difference in Dawley rats combined (discussed laboratory (G-114 at 21).This factor led tumor incidence of approximately 13% above). These tumor findings suggest a Dr. Hicks to conclude that the total between the controls and the high dose low treatment effect and thus emphaslzo tumors found in this study were [2.5gm/kg) treated animals (G-41, App. the need for studies of greator "pathologically * * * very significant" Vat 19). This study is therefore unlikely sensitivity than the Taylor study. (id. at 20).Moreover, as previously . to detect a small treatment effect. This Accordingly, I cannot consider this noted, Dr. Brown testified that the lack of sensitivity is especially study to be proof of cyclamate's safety. "probability of observing three or more important in view of the findings of (4)Hamberger et 01. (A-340). (a) Study tumors in the 217 treated animals in [the bladder tumors in Sprague-Dawley rats Design: The Hornberger study involved Hicks and Ikeda] studies combined is combined and Wistar rats combined groups of 25 Charles River CJ)-1 .OOZ" (assuming a background rate of (discussed above). Accordingly, r cannot Sprague-Dawley male rats which were .116% tumor incidence as obtained from consider this study to be proof of fed O. 1 or 5% sodium cyclamate. The the NCI data on all species of rats' cyclamate's safety. . bladders of at least 12 animals per group combined) (G-120 at 11). Although this (3) Taylor, et 01. fG-13}. (a) Study were examined microscopically. They evidence does not conclusively establish . Design: The Taylor study involved 48 were started on test at approximatoly that cyclamate is a carcinogen, the study mall! and 48 female Sprague-Dawley six weeks of age and continued on cannot be considered proof of safety strain rats fed a diet containing 5% treatment for two years. and indeed raises a question as to the calcium cyclamate. The animals were (b) Study Results: The authors of tho potential carcinogenicity of cyclamate. deriveeI from parents who were also study concluded that: (2)Ikeda, et 01. fG-79}. (a) Study administered cyclamate from the time of On the basis of these experiments, 1\ Design: This study involved groups of mating through delivery and weaning of cannot be concludedthat ' • • cyclamata 54-56 male Wistar rats fed sodium the test generation. The study was [is]carcinogenic. This may be consideredof cyclamate or a sodium saccharin plus continued for 114 weeks. significance sincefor smallerdoses of other sodium cyclamate mixture. The' (b) StudyResults.' One bladder tumor compounds under similarconditions were concentration given was 2%for the first was found in a control animal and none unquestionably carcinogenic for liver. Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61493 bladder. subcutaneous, vascular and other As to the possibility ofcross to comment on the apparent failure of tissues of rats and/or mice. contamination of the animals In the the Schmaehl study to report the results (A-348 at 9J.Two carcinomas of the Homberger study with other chemicals. of the study separately by sex, Abbott bladder were found in cyclamate including a bladder carcinogen. I find contends that a number ofeffects were treated animals (one in the high dose that the likelihood of cross reported by sex and that even ifthe group and one in the low dose groupJ contamination is too speculative to be reported Incidences of tumors all ~d none in control animals (Tr. at 602 relied upon, especially where I have occurred in one sex, none of the 604J.The ALJ found that the Homberger found no tumors occurring in control reported findings would be statistically study employed an "unacceptably small animals (see Section IV.B.2.b.(l)(c) slznlficant at P,.05 (A-858 at 23). number of animals per group" (ID at 31J, below). (Findings other than tumor finding~, (c) Analysis: Abbott takes exception (5)Schmaehl (A-555). (a) Study such as waterintake and body welght to the ALI's finding (Abbott's Exceptions Design: This study involved groups of gains, were reported by Dr. Schmaehl at 29). Abbott, relying on the auth01'l!' 104 Sprague-Dawley rats fed either • \..ith information about the sex of the conclusion, contends that the study IS sodium cyclamate. sodium cyclamate animals. However, this information is negative (Abbott's Brief at 17, 22). and saccharin, or CHA. Animals were not relevant to the question raised by The Bureau relies on the Temporary started on study between 70-90 days (1~ the Remand Order. That question was Committee Report. The Report stated age and continued on treatment for their intended to inquire whether certain that: lifetimes. ' tumors may have been statiscaJly (b) Sw.dy Results: One bladder tumor [a] number of questions were raised significant ifthey occurred only in one was found in a cyclamate treated regarding the experimental design ~d sex) conduct of this study. The smaIl animal group animal. The authors of the study The Bureau argues that tumor findings size and the possibility of cross concluded, however, that "[iJn spite of statistically significant at P,.05 may be contamination of the cyclamate-treated the high dosages and the duration of the present in the Schmaehl study because, animals with other chemicals being tested in experiments over an entire ~etim:. no the same room, including .onelaterfound to if the tumors of the reticuloendothelial evidence was found of chromc tOXlC or system (reticular cell sarcomas, be a bladder carcinogen, limit the value of carcinogenic activity of the substances this study in assessing the carcinogenicity of lymphosarcomas and leukemia cyclamate. tested" (A-555 at 6). The Temporary combined) all occurred in the same sex, Committee found that the reported their incidence would infact be (G-41 at18; Bureau's Brief at 14). The extremely rare occurrence of Bureau concludes that the Homberger statistically significant at P<.05 when spontaneous bladder tumors in the rat compared to controls (Bureau's Remand study is inconclusive (Bureau's Brief at strain used "must be taken into 19J. Brief at5-0). consideration when evaluating the I agree with Abbott that the incidence I agree with the Bureau and the ALJ significance of the one bladder that the sample sizes employed in the of lymphosarcomas, reticular cell transitional cell carcinoma found in a sarcomas, and leukemias occurring in Homberger study were unacceptably cyclamate-treated animal" (G-41 at 23 small. The Homberger study had only a the Schmaehl study, ifexamined 24). independently, would not be 50%chance of detecting. at the·95% _ (e) Analysis: Abbott, rel,1,iD8 on the statistically significant. even ifoccurring confidence level, a true difference in conclusion of the authors of the study. tumor incidence of approximately 26% in one sex (R. Tr. at 215].I find, contends that tha Schmaehlstudy u however, that by comblnlng eltIte.r between the controls and the highdose negative (Abbott's Briefat 23-24). The reticulum cell sarcomas and (5%J treated animals (G-41. App, V, at Bureau does not criticize the conduclor 19). This study therefore is unlikely to lymphosarcomas qr these twa effects design of this study, but considers the and leukemias, a result statistically detect a low treatment effect. This lack in one bladder tumor found thisstudy significant at P..;.05wouldbeachieved, of sensitivity is especially importantin together with the tumors found in the view of the findings of bladder tumors in if these effects occurred all in one se-x. Homberger and Taylor studies and (G-140 at 3-4: R.Tr. at 161; 21l}-11; 215 cyclamate-treated rats in this study (G contends the results are biologically 121 at 7) and in Sprague-Dawley rats 16]. Such a combination ofthe data is important. The Bureau's contention appropriate because lymphosarcom!1s, combined and Wistar rats combined concerning the analysis of this study (discussed above). Accordingly, the reticular cell sarcomas, and leukemias with other similar rat studies and all involve cells derived from reticulum study cannot be considered proof of Abbott's exception to this analysis is cells [R. Tr. at113-19). Without a report cyclamate's safety. discussed above. As noted there, the The author's remark that smaller of these tumor findlngs by sex, this Issue occurrence of a bladder tumor in the cannot be conclusively resolved. doses of 'Othercompounds under similar Schmaehl study is consistent with a conditions were unquestionably small treatment effect (G-12O al6-7: see Abbott's position is that a detailed carcinogenic under similar conditions G-126 at 11-12), even though it is not report of the tumor findings by sex is does not alter my conclusion. At best, significant at the P".051evel. Moreover. nevertheless unnecessary. Abbott that finding only tends to sho,:" that . as the Temporary Committee noted. this contends that a statement contained in cyclamate is not a strong carcinogen in finding must be viewed in light of the the Schmaehl report and a conversation this species. However, because of the extremely rare occurrence of between Dr. Oser and Dr. Schmaehl are poor sensitivity of the study, it does not spontaneous bladder tumors in this rat sufficient to resolve this issue (Abbott's provide any reliable insight into whether strain (G-41 at 23-24). Accordingly. I Remand Bx,at 21). The Schmaehlrepvrt cyclamate is a weak carcinogen.P cannot consider the Schmuehl study to states that "No greater incidence be proof of cyclamate's safety. regardlng either sex could be detected :3The terms "strong" and "weak" carcinogen an! (d) Other matters: As part of the wIthreference to the benign or the used here to differentiate between compounds Remand Order, the parties were asked malignant tumors" (A-555 at 5J. which respectively cause relatively high and . Additionally, in a conversation with Dr. relatively low incidences of tumors when,tested.";1 Oser, Dr. Schmaehl is alleged to have experimental animals. Even a "weak ~ogen , populatlon as a ",bole-Indeed.evena 1°~ ini:fi:,;:e however. by this distinction can cause ~portdnt in tumor incidenoe would be unacceptable In the said that he would have reported and unacceptable incidences of cancer in the human human population (Tr. lit 102). significant differences as to sexifthey 61494 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices
were present (A-858 at 20).It is unclear, cyclamate direct feeding studies and independently, were published together however, whether Dr. SchmaehJ's found that: and are discussed below. statements refer solely to specific tumor In the rat studies. seyen transitional cell (1)Friedman, et al. (A.-38B). (a) Study findings, e.g; lymphosarcomas, or carcinomas of the bladder, two of the kidney, Design: The first of the two Friedman whether they also refer to combined three bladder papillomas. five hyperplastic studies [hereafter "first Friedman tumor findings, e.g; effects on the entire lesions, and a bladder proliferative lesion study") was conducted using seven male reticuloendothelial system. This were found in rats treated solely with and seven female Osborne-Mendel rats question is particularly important in cyclamate [Hicks. (Ex.No. G-2). Plank (Ex. per group. These rats were fed sodium view of Abbott's questioning of the No. A-146). Friedman (Ex. No. A-195) and cyclamate or calciumcyclamate at 0.4%, propriety of combining effects on the Schmaeht(Ex. No. A-0555) [Studies]]. 2.0%,or10% of their chow diet for 101 reticuloendothelial system (R. Tr. at 117 weeks. A group of 14 controls per sex 18). IfDr. Schmaehl shared Abbott's (ID at 31). fed a-standard chow diet was skepticism about the combining of . (c) Analysis: Abbott takes exception maintained. The animals were started as effects, he probably would not have to this study's being grouped with other weanlings and the study continued for analyzed combined effects on the studies involving rats fed cyclamate 101 weeks. reticuloendothelial system. Accordingly, rather than CHA (Abbott's Exceptions The "second Friedman study" was without a report of lymphosarcomas, at 27). Abbott also contends that in . conducted using male Holtzman rats. A reticular cell sarcomas and leukemias grouping the Plank Study with other' group of twenty of these rats were fed a by sex, the 'precise meaning of Dr. studies (1) the ALI erroneously grouped semisynthetic diet containing calcium Schmaehl's statements remain together carcinomas and cyclamate at 1% level plus 20%casein, uncertain. As noted previously, it is not noncarcinomas, such as papillomas, and and 2%level plus 20%casein. and 2% unusual for scientists to disagree with (2) that the three bladder papillomas level plus 10% casein. An equal number the conclusions 'of the author of a study cited by the ALI were not confirmed by of controls were fed the semisynthetic as to the significance of the results of a " the Pathology Working Group (Abbott's diet with 20% casein. study. I therefore agree with the ALI's Exceptions at 27). The Bureau does not .(b) Study Results: Three transitional conclusion that "The only conclusion dispute these points (Bureau's Reply at cell carcinomas (two at the low dose that can be drawn from the author's 14-15). I agree with Abbott that a study: and one at the high dose) and two failure to report this data separately by of CHA should not be grouped with. papillomas of the bladder were found in sex is that it is uncertain whether a true cyclamate studies, although it can, by the calcium cyclamate treated unlmnls sex specific effect occurred" (IRD at12). itself, provide important Information in the flrst Friedman study (ID at 11). It,should be noted that Dr. Oser's about the safety of cyclamate because it Three papillomas were found in the conversation with Dr. Schmaehl was is a metabolite of cyclamate in humans. sodium cyclamate treated rats in the stricken as hearsay and Abbott took I also agree '¢th Abbott that the ALI first Friedman study [id.). One paplllomn exception to this ruling. Although I agree erroneously lumped noncarcinomas was found in a calcium cyclamate with the ALI's ruling, I have together with carcinomas. I find, treated animal in the second Friedman nevertheless considered the statement however, that the Plank study does not study (G-:41,App VII,Food and Drug and found that it does not resolve the prove the safety of CHAo The sensitivity . Administration L. Friedman et al. 1972 at issue because of the ambiguity of the Plank study is unacceptably low. 4). contained in the statement. The Plank study had only a 50%chance . (6) Plank, et 01. (A-401-404). (a) Study The Pathology Working Group of tho of detecting, at the 95% confidence level, Design: This study involved Charles . Temporary Committee confirmed the a true difference in tumor incidence of River CD-l Sprague-Dawley albino rats, . three bladder carcinomas found in the . approximately 33% between the controls in groups of 25 of each sex, fed the • calcium cyclamate treated animals in and the high dose (15 mg/kg) treated the first Frfedman study, but did not following concentration'of (~1, cyclohexylamine sulfate: 0.15 mg/kg/ animals App. Vat 20). As a confirmthe papillomas. The Temporary result, statistical slgnlflcance at the ' day, 1.5 mg/kg/day, or 15 mg/kg/day. A Committee found that: : control group of -25 of each sex was also P<;;;.05Ievel is difficult to demonstrate , The small number ofrats used is maintained. The study was conducted unless the test substance causes an considered to be a major deficiency in thls for two years, after which the survivors exceptionally high tumor incidence (G study. Although the incidence of bladder were sacrificed. • 121 at 8). The single bladder tumor tumors was not statisticaUy significant, their (b) Study Results: A single, bladder ' found in a cyclohexylamine treated importance. even in small numbers, must bo carcinoma was found in one male from animal has biological significance evaluated with respect to tho reporled rarity because the occurrence of bladdej, of spontaneous bladder tumors in the rat the high dose treatment group (G-41 at strain used. 24). The Temporary Committee found tumors in the strain of rats employed in . that "[tjhe value of this study is limited the Plank study is rare. This single (G-41 at 20-21). The calcium cyclamate by its poor sensitivity. It is thus .bladder tumor may be an indication of a portion of the first Friedman study WM considered to be of minimal value in weak carcinogenic effect which might among the studies that the Temporary assessing the carcinogenicity of have been statistically significant if the Committee found create a "sense of cyclohexylamine"(id). The ALI found study had been larger. The single uncertainty" about the safety of that "[b[ecause of the extreme rarity of bladder tumor found in the Plank study cyclamate (id. at 46). spontaneous bladder tumors in this thus has blological significance (id. at 8 The ALI, who recognized the strain, the positive flnding raises 9). Accordingly, I cannot consider this deficiencies in the first Friedman study questions concerning CHA's study to be proof of cyclamate's safety. cited by Abbott (discussed below), carcinogenic potential. Furthermore, the b. The Occurrence ofBladder Tumors concluded that with respect to the first study's sensitivity was limited due to in Holtzman and Osborne-Mendel Rats: Friedman study the "incidence of tumors the small number of animals used" (ID . Two studies submitted in the cyclamate is important, even though not at 16). hearing were conducted by Friedman, et statistically significant. because The ALI also grouped the Plank study al. and involved Osborne-Mendel or spontaneous tumors are extremely raro with the Hicks, Friedman, and Schmaehl Holtzman rats. These studies conducted in the rat strains employed" [ID at 12). Federal Register I Vol. 45. No. 181 I Tuesday, September 16. 1980 I Notices 61495
(c) Analysis ofthe Calcium If there were some correlation between utilized a small number of animals. Cyclamate Portion ofthe First Friedman the increased dose levels and an Although this portion of the Friedman Study: Abbott contends that (1) the three Increase in tumor production. I would study utilized a small number of carcinogens found in calcium cyclamate have greater confidence in the results of animals, the small sampler size is not a treated animals in the-first Friedman the study. However. the lack of such a valid reason for discounting the three study were neither statistically response may have been due to the bladder tumors found in the calcium significant nor dose related; (2) the small small sample size (G-12O at 13). A small cyclamate-treated animals. A study, number of rats used in this portion of the sample size makes the finding of a dose such as the Friedman study, which first Friedman study is a major response more-difficult, because of because of its insensitivity is unlikely to deficiency; (3) the tumor findings are random fluctuation (id). detect a carcinogenic effect. may complicated because they appeared only I reject Abbott's argument that the nevertheless detect a carcinogenic effect in the first Friedman study which results of this portion of the first in some cases. There is nothing utilized a chow diet and did not appear Friedman study are unreliable because inconsistent in finding that a study is too in the second Friedman study which tumors appeared only in animals on 11 small to yield reliable negative results utilized a semisynthetic diet; and (4) the chow diet (used in the first Friedman yet is sufficiently sensitive to raise first Friedman study is complicated by study) but did not appear in animals on serious doubts as to the safety of the the presence of calculi and bladder a semisynthetic diet (used in the second tested substance (see Tr. at 630-31). parasites (Abbott's Brief at 23). Friedman study). Abbott's argument Thus. the lack of sensitivity of the The Bureau contends that the three might have merit but for the fact that all sodium cyclamate portion of the bladder tumors found in the calcium control animals in the first Friedman Friedman study is not a valid reason to cyclamate portion of the first Friedman study received the same chow diet criticize the finding of three bladder study are biologically significant, (absent cyclamate) as the treated tumnors in the calcium cyclamate notwithstanding the lack of statistical animals and there were no tumors found portion of the study, even though both significance at the P<;.051evel. because in the control animals. The use of portions of that study employed the the spontaneous bladder cancers in mice concurrent controls in which no tumors same number ofanimals. and rats are rare (G-121 at 8). The were found negates the possibility that Finally, Abbott contends that the Bureau further contends that the lack of tumors found in treated animals were three tumors in the calcium.cyclamate dose response might be attributable to due to the chow diet (see Tr. at 1049-50). treated group may have been due to the small size of the study [Bureau's Thus, the study design ensured that the bladder calculi::· or bladder parasites. Brief at 24). results of the study would not be biased The evidence on the relationship. I find that the three bladder tumors by the type of diet received by the between bladder calculi and tumors is found in the calcium cyclamate treated cyclamate treated animals. at present inconclusive (G-41, App V at 48-49). Moreover. in a related context, animals of the first FrieWnan study add Moreover, Abbott's only citation for to the doubt about the safety of Abbott contends that "ifa study is to this contention is the report of the study cyclamate that was raised by the have relevance onwhether parasites findings of bladder tumors in Sprague (A-195;Abbott's Brief at 18).This cause bladder tumors. the length of Dawley and Wistar rats discussed in reference does not state that the results exposure to parasites must be known" subsection B.2.a. above. of the first Friedman study are (Abbotfs Exceptions at 31).This • I recognize that the three bladder complicated by the chow diet. but rather commentwouldseem.to apply equally tumors found in this study were not describes the results of the to bladder calculi. Abbotthas notcited significant at the P<;.05 level (P=O.29; G histopathology for the two studies (A any evidence as to each animal's length 41,App. VII,Food and Drug 195 at 755-56).The only other support of exposure to bladder calculi or bladder Administration Friedman et al. at 2). for Abbott's contention that could be parasites. It may be that the bladder These tumors are nevertheless found is the report of the Statistics tumors in this portionofthe first biologically significant because (1) the Working Group to the Temporary Friedman study were causedby bladder sensitivity of this portion of the Committee which speculates that the calculi. However. the evidence Friedman study is low (this portion of tumors in animals on a chow diet "may" submitted is insufficient to establish that the study had only a 50% chance of have been due to contamination (G-41. the bladder tumors were causeby detecting, at the 95% confidence level. a App. VII, Food and Drug Administration bladder calculi. true difference in tumor incidence of L Friedman et al. 1972 at 4). However. Since the randomly selected control approximately 34%between the controls there is not evidence or other group presumably had an equal chance and the high dose treated animals (G-41. explanation supporting the suggestion to develop such calculi..the observed App. V at 19)) and (2) the spontaneous that the chow diet may have been bladder calculi may be treatment related or background rate for bladder tumors in contaminated. Moreover. the report of in which case cyclamate might be Osborne-Mendel rats is reported as the fullTemporary Committee didnot producing a carcinogenic response. being low (G-41 at 20-21; G-121 at 8). state that the chow diet was a albel t an indirect one. Thus. evenif Thus, the occurrence of three bladder complicating factor and recognized the there were definitive evidence that the tumors in the calcium cyclamate treated potential importance of the three bladder tumors in this study were rats is consistent with a small treatment bladder tumors found in the calcium caused by bladder calculi (which there effect even though they are not cyclamate portion of the first Friedman is not), itwould not resolve the question significant at the P<;.051evel.The similar study (G-41 at 20-21). I therefore of the safely of cyclamate. findings of bladder tumors in Sprague conclude that there is no llasis upon In sum. the three bladder tumors Dawley and W"lStar rats reinforces the which to attribute the three bladder found in the calcium cyclamateportion conclusion that these three bladder tumors found in the cyclamate treated of the first Friedman stW1y do not tumors are biologically significant animals in the first Friedman study to conclusively establish that cyclamate is The calcium cyclamate portion of the the chow diet a carcinogen. Moreover. this finding first Friedman study is, however. further I also reject Abbott's argument that questioned by Abbott because there the calcium cyclamate portion of the :. Ca1aill are CllllCE.."WOII& lI5UlI1Ir oCJDineral salts was no clear dose response relationship. Friedman study is deficient in that it lIlOW1 does not provide the same degree of instilled in their bladders [Hicks. A-832,· which they were sacrificcd and givcn a confidence that one would have if the G-2): and (3) the direct feeding of a histologic exam. results of the study were significant at . cyclamate/saccharin mixture to rats (2) Study Results: The ALJdescribed the P<.05Ievel. This bladder tumor (Oser, G-81). It is undisputed that the the results of the Bryan study as follows: finding does. however, add to the doubt incidence of bladder tumors in the • • • the incidence of bladder tumors In raised by the bladder tumors found in treated group in all three of these the animals Implanted with cholcsterol and Sprague-Dawley and Wistar rats. studies is statistically significant. cyclamate pellets was 78~G whereas tho (d) Analysis ofthe Sodium Cyclamate Abbott argues that. even ifproperly incidence was 13%'in the controls. IQn . Portion ofthe First Friedman Study: The conducted. the techniques employed in duplicate experiment conducted by Dr. Brynn Bureau contends that the sodium . the Hicks and Bryan study are invalid the incidence was 61% In the test animals and cyclamate portion of the first Friedman - for assessing the carcinoginicity of a 12%in the controls. ,.studyshould not be given any weight as substance. Abbottfurther argues that A positive control group was also a negative study because of the small the presence of saccharin in the Oser' maintained. Mice were Implanted with number (fifteen) of animals treated cholesterol pellets containing a-methyl ethlJr study makes that study inappropriate for xanthurenlc acid, a compound prevIously (Bureau's Brief at 24). The Bureau also assessing the carcinogenicity of found to be carcinogenic in mouse bladders, notes that the dose levels were rather cyclamate. The incidence of bladder tumors was 35%. till low in this portion of the first Friedman Although these studies are not as expected. • study (G-120 at 12). reliable as direct feeding studies, such ,I agree with the Bureau. The size of (ID at 18).The ALJconcluded that: as the Rudali study, Hind that the u. .. •although there are questions as tliis study is too small to permit reliable results of these studies give rise to a conclusions concerning the safety of to whether [the Bryan technique] is still high degree of suspicion concerning the an appropriate procedure ••• the results cyclamate. This portion of the first possible carcinogenicity of cyclamate Friedman study had only a 50% chance cannot be totally disregarded. The and add support to the bladder tumor results represent a major biological and of detecting, at the 95% confidence level. . findings in the cyclamate direct feeding a true difference in tumor incidence of statistically significant effect which has studies discussed in Section IV.B.2. I not been satisfactorily explained and approximately 34% between the controls recognize that the significance for and the high dose (10%) treated animals. which increases doubt concerning human health of the findings in the cyclamate's safety" [ID at 32), (G-41, App. V at 19).This degree of , Hicks and Bryan studies can not yet be sensitivity is unacceptably low (G-120 The Temporary Committee. however, fully evaluated. We do know. however, concluded that "the rout of . at 12).This portion of the study is that the difference between a low therefore too insensitive to be administration [in the Bryan study] is considered proof of safety. incidence of cancer and no incidence of inappropriate for assessing the (e) Analysis ofth'1 SecondFriedman cancer (as in the Hicks controls) is the carcinogenicity of a human dietary Study: 1find that the size of the second presence or absence of cyclamate (G- . constituent" (G-41 at 27).The BryalJ, Friedman study is too small to permit 112 at 20).The suspicions raised by study was nevertheless among those reliable conclusions concerning the these studies could be negated by valid studies referred to by the Temporary safety of cyclamate. The study had only and convincing negative direct feeding Committee which created a "sensa of a 50% chance of detecting at the 95% studies or evidence that the uncertainty" about the safety of confidence level. a true difference in carcinogenic response is unique to this cyclamate (G-41 at 46). tumor incidence of approximately 20.5% mode of administration and could not (3)Analysis: Abbott takes exception between the controls and the high dose result-from ingestion of cyclamate. to the ALl's finding with respect to the (2%) treated animals (G-41, App. V at However, as the dlscusslon in Sections Bryan study (Abbott's Exceptions at 24). 19).This degree of sensltivity is IV.C. and D. establishes. Abbott has Abbott contends that the Bryan study unacceptably low. This study is failed to submit such studies. contributes nothing to the evaluation of therefore too insensitive to be , 1note that these three studies do not the carcinogenicity of cyclamate considered proof of safety. playa major role in my decision. Indeed, (Abbott's Brief at 32).Abbott reHes on (fJ Other matters. As part of the even in the absence of-these studies, 1 the testimony of Dr. Bryan who stated Remand Order.' the parties were'asked would reach the same conclusion. i.e., that ". • • It's a technique that is not at to comment on the reported increased that cyclamate has not been shown to all replicative of normal human overall mortality in the Friedman study be safe. 1have, however. given these experience, experimental variables may and the author's report that a small studies some weight because. although be difficult to control " • 11 [and] , number of animals in the study were the Hicks and Bryan techniques andthe utilizatlon of this technique really has unaccounted for.'The parties stipulated Oser.study may not involve the methods diminished substantially in the last that these events result in a smaller pool of choice and should not be relied on several years " • ." (G-113 at 17-18: of animals from which to measure primarily to screen food additives for see also G-120 at 16). biological effects and that therefore, the carcinogenicity, these methods have The Bureau contends that the Bryan ability of the study to detect biological shown biological effects cannot be sfiidy is positive [Bureau's Briof at 19). effects is decreased (Stipulation dated ignored (G-112 at 20).The scientific The Bureau notes that Dr. Bryan September 17,1979at 3). . basis for this conclusion is discussed conducted two replicate experiments 3. The occurrence ofBladder Tumors below. one year apart. Both of the cyclamate In Rats In Studies Other than a. Bryan, et al. fG-I). (1) Study treated replicate groups developed Cyclamate or Cyclohexylamine Direct Design: The ALJ described the Bryan augmented incidences of bladder tumors Feeding Studies. The three studies study as follows: when compared to the respective control discussed in detall below involve (1) the Cholesterol pellets containing 20%sodium groups. In both instances the statistical implantation of a pellet consisting of cyclamate were surgically implanted in the evaluation revealed a highly significant cyclamate and cholesterol in the . bladders of 100 female swiss mice. A control difference between treated and control bladders of mice (Bryan. G-l): (2) the group of 100 mice with cholesterol pellets in groups (G-113 at 15-16). direct feeding of cyclamate to animals their bladders was maintained. The mice I find that the Bryan study does which have a potent carcinogen (MNU) were permitted to survive 55 weeks after support the conclusion that cyclamate Federal Register I Vol. 45, No. 181 I Tuesday, September 16. 1980 I Notices 61497 has not been shown to be safe. I study. The Bureau argues that the latter effect was statistically significant recognize that the method used raises presence of a foreign body in the urinary (p<.05). the possibility that the carcinogenic tract is a condition that occurs in human (4)Requestfor Rebuttal: Abbott also effect seen may be caused in part by the pathology (G-113 at 19). The Bureau contends that rebuttal testimony it instillation technique or the cholesterol further argues that surgical procedures attempted to submit on the Bryan pellet or both. However, I disagree with on the bladder do occur in people, some technJque was wrongfully excluded. I the Temporary Committee's finding that of whom might be exposed to cyclamate disagree. The purpose of rebuttal the route of administration in the Bryan both before and after surgery (Tr. at testimony is to allow the party with the study was inappropriate because the 818). Finally, the Bureau argues that burden of proof to adduce evidence on animals which were exposed to the stone formation can occur after a matters the relevance or existence of cyclamate incorporated in a cholesterol surgical procedure is performed on the which were unknown or could not be pellet and surgically implanted in the bladder (id.).l do not find these reasonably foreseen at the time ofthe animals' bladder were compared to arguments totally convincing. The presentation ofits case in chief. Abbott control animals exposed to the same tumors found in cyclamate treated conceded that it "has long been familiar type of surgically implanted cholesterol animals may have been due in part to with the Bryan study" (Amended Motion pellet. The only difference between the the unique circumstances of this test. for Leave to Adduce Rebuttal Testimony treatment and control groups is the Abbott's unsupported argument that at 8). Abbott nevertheless attempted to exposure to cyclamate (G-112 at 20). these unique circumstances are justify its rebuttal testimony on the The statistically significant difference . responsible for the tumors found in ground that it did not anticipate that the between treatment and control thus cyclamate treated animals in the Bryan Bureau would rely heavily on the Bryan shows that cyclamate is the sole or the study is, however, insufficient by itself study [id. at 8). This reason is primary cause of this tumor production. to rebut the suspicion raised by the insufficient. Ifallowed. it could permit I do not find Abbott's attempts to Bryan study.l find that the Bryan rebuttal testimony on almost any topic. explain these tumor findings, which are technique is sufficiently analogous to The Bryan study is one of eight studies discussed below, persuasive. I agree human experience to require that valid that the Bureau characterized as with the ALJ that the Bryan study has and convincing negative direct feeding positive and does not appear to have shown "a major biological effect" which studies (using the conventional route of been given any more reliance than the adds to the doubt concerning administration) or studies which prove other seven studies. Abbott thus had cyclamate's safety. that the carcinogenic response is unique ample opportunity to submit testimony Although Dr. Bryan admitted that the to this technique and will not occur as a challenging the Bryan study, and • technique he used is not replicative of result of direct feeding of cyclamate. be therefore was properly precluded from human experience and that its use has submitted to rebut the suspicion raised submitting rebuttal testimony on this declined substantially in the last several by the Bryan study. issue. years, Dr. Bryan did testify that the The Bureau also argues that the total Moreover. I agree with the ALJ that reliability of the technique is supported duration of exposure of the bladder to the proposed rebuttal testimony of Dr, by concordance of results, both positive cyclamate in the Bryan study is less Clayson [which would have argued that and negative, between laboratories than a day. whereas human exposure to the Bryan technique was invalid and where careful studies have been cyclamate as a food additive. while involved conditions which do not occur conducted (G-113 at 19). Dr. Bryan involving considerably lower levels per in humans) is "largely in the nature of further testified that the correlation day, would involve exposure for a much argument and should best be presented between the results of studies in which larger period of time (25.000days in a as argument on brief' (Order of the pellet implantation technique is lifetime) (Bureau's Brief at 39; G-112 at September 12. 1977 at 1). These employed and those utilizing direct 18-19). The Bureau concludes that arguments were presented by Abbott in feeding studies is "remarkably high" (Tr. because of the longer exposure, the their brief and are fully considered at 828).Finally, Dr. Bryan testified that carcinogenic effect seen in the Bryan above. his technique has validity where large study "could be potentiated many, many Abbott also argues that the ALJ erred enough population samples and fold in an exposure continued for in refusing to allow it to submit a adequate controls are utilized (Tr. at thousands of days in a human published scientific journal article by 823). population" (Bureau's Brief at 39). I am [ull, et al. (A-853). The procedural It is also important to note that the not persuaded by this argument. The regulations governing the submission of bladder implantation technique is animals in the Bryan study arc exposed such articles required Abbott to submit currently being used by three other to a single brief (short elution time) but all documentary data and information groups, which suggests the continuing intense and highly localized exposure. to upon which it sought to rely by June 15, vitality of the technique (Tr. at 823). the unmetabolized agent, directly at a 1977 [21 CFR 12.85(b)). The Jull article Moreover, the usage of different target site. Although human exposure to was published in 1975, but was not techniques is not due to a lack of cyclamate ingestion would be long term, offered into evidence until November 3, confidence in the Bryan technique, but it would also be systemic exposure with 1977. some five months after the date for rather is largely due to the ability of relatively lower concentrations at any its submission and two months after different techniques to generate bladder given tissue. Thus, because of the cross-examination was completed. The tumors more quickly, in a higber yield, differences in the nature of the procedural regulations allow a and with less expense than the Bryan exposure. the longer term of the human participant to supplement its submission technique (G-113 at 18). exposure would not necessarily result in under § 12.85 where "the material It is possible that the surgical .a greater carcinogenic effect in humans contained in the supplement was not procedure used to implant the pellet, or than was found in the animals in the reasonably known or available when the implanted pellet itself, acted Bryan experiment. It is, however. the submission was made" 21 CFR synergistically with cyclamate to unnecessary for the Bureau to show that 12.85[c). Abbott claimed that in spite of produce the tumors found in the the effect in humans would be greater its efforts to locate this information. the cyclamate treated animals in the Bryan than that shown in animals. because the article was unknown to it prior to 61498 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices ,,- October 31. 1977. "primarily because the 20 percent tumor rate, but the MNU plus (i) Relevance to Human Experience: study was not published under its own cyclamate group again produced a 50 The theory underlying Dr. Hicks' title. but rather within the book whose 'percent incidence of bladder cancer (G technique is important toan title gives no specific indication of its 64). The results in both of these understanding of its relevance to the existence." (Motion to Add A Document experiments were highly statistically human experience. Dr. Hicks utilizes to the Administrative Evidentiary significant (p<.001) (G-41, App. VII). MNU as part of her technique because Record at 3). The ALI ruled that this The Temporary Committee found that there is a very good dose response reason is insufficient. I agree. In view of the "MNU-plus-cyclamilte regime relationship between MNU and the the fact that this article 'Wasnot offered resulted in a clear carcinogenic response incidence of bladder cancer (G-114 at 8). until five months after it was due, two in the treated animals" (G-41 at 25). The A single intravesicular dosage of either months after the completion of cross Temporary Committee further found that 1.5 or 2. mg of MNUhas been shown to examination and one month after the Hicks technique "may very well be "noncarcinogenic" in the bladder (G AbboU's request for a ruling on rebuttal become an important screening method 2 at 226;Tr, at 991).A second similar testimony and that this article was r for substances suspected of being a dosage (at either 1.5 or 2 mg] of MNU available two years before the date for urinary bladder carcinogen" but that will cause tumors (G-2 at 226-27). The submission of such articles, 1find that it "[i]t has not yet ••• been validated for Hicks method involves substituting the was properly excluded from the this purpose" (id. at 45-46). The Hicks test substance for a second dose of the evidentiary record. - study was among the studies that the _ known carcinogen MNU. The underlying Although the [ull article and the Temporary, Committee found "create a theory of the Hicks method is that if the proposed testimony ofpr~Claysonwere sense of uncertainty" about the safety of test substance docs produce tumors, it is • properly excluded, I have nevertheless cyclamate (id). either initiating the tumor production decided to consider the main argument The ALJ found that the Hicks study and thus is a carcinogen or is promoting contained in these submissions. The raised serious questions concerning the the effect'by acting synergistically with Bryan technique is criticized on the carcinogenicity of cyclamate (ID at 20, the MNU (G-114 at 8). The Hicks ground that if the control animals are 32). The AL} found further, however, methodology was explicity or implicitly kept alive for their normal lifespan the that several factors raise questions as to endorsed by five leading oncologists or tumor incidence is so high that valid the validity of the Hicks study: (1) the toxicologists (G-113 at 9-10; G-118 at 11; conclusions concerning the feed was not analyzed for pesticides; (2) G-112 at 17; G-121 at 9; Tr. at 1173). carcinogenicity of a test compound no separate control group was The only difference between tho cannot be drawn, This criticism lacks anesthetized or instilled with an ' treatment and control groups in the merit. Dr. Bryan testified that he had innocuous material; and (3) the lack of a Hicks studies was the feeding of conducted an experiment (unrelated to formal randomizationmight have cyclamate (G-112 at 20). Thus. it is his cyclamate experiment) in which a introduced additional bias (id. at 20). reasonable to attribute the high tumor (3) Analysis: Abbott takes exception cholesterol pellet was left in the mouse incidence found in the cyclamate treated to the ALl's finding that the Hicks study bladder for 110 weeks (the normal raises serious questions concerning animals to cyclamate. Although the lifespan of a mouse) and found that the cyclamate's carcinogenicity. Abbott precise mechanism by which this tumor incidence of bladder tumors under that _contends that (1) the technique used by incidence was caused is unknown, a circumstance is only about 12, 13, or 14 Dr. Hicks is unlike any human convincing explanation has not been percent (Tr. at 803). Indeed, the [ull circumstance and therefore suspect; (2) provided as to why these results, which article recognized that the high tumor Dr. Hicks was uncertain as to the are highly statistically significant. incidence lull found in the strain of mice technique she actually used; (3) attempts cannot be attributed solely to cyclamate. he subjected to the Bryan technique and to replicate her work have failed; (4) Moreover, the fact that a known kept alive for a normal lifespan could be' there were no anesthetized controls or -carcinogen, MNU, would produce a , unique to the strain of mice used in the controls instilled with innocuous similar increase in tumor producllon if lull study and might be different for materials: (5) no formal randomlzation substituted for cyclamate in the Hicks other strains (A-853 at 388-89). This was used: and (6) the 'animal facilities model supports the conclusion that strain variation could explain why Dr. and environmental control were below cyclamate is producing a carcinogenic Bryan found a 12, 13 or 14% tumor the optimum standards. Abbott further effect under the circumstances of this incidence. in the mice he 'kept alive for contends that the AL} erred in refusing test model. I therefore find that it is' 110 weeks. Thus, the [ull article does not to allow Dr. Deutsch Wenzel to-appear reasonable to attribute the tumors rebut Dr. Bryan's findings. and testify in rebuttal (Abbott's produced in the Hicks study to b. Hicks, et al. (A-832). (1) Study , Exceptions at 22). cyclamate. Design: This study involved rats whose 'QIe Bureau takes exception to the Abbott contends that the instillation bladders were stripped of the epithelium ALl's criticisms of the Hicks study. of MNU in the bladder of tho test animal by instillation of methylnitrosourea contending that (1) although Dr. Hicks is unlike any human experience and (MNU), a potent carcinogen, and then did not use a table of randomization. she thus renders the Hicks model totally fed sodium cyclamate. A control group, did use an appropriate system of inappropriate. In support ofthls given an MNUinjection was also randomization: (2) analyzing feed for contention, Abbott notes that "the basis maintained. , pesticides is unnecessary for such a' of the (Hicks) model Is to Initiate (2) Study Results: When animals were study and, in any event, both treated neoplastic changes with MNU" and that exposed to'MNU and fed a cyclamate and control animals received in the '~in focal areas, the eplthellumIs containing diet for two years, a very same feed; and (3) there is no evidence stripped" (Abbott's Exceptions at 23). high incidence of tumors developed in that the anesthetization or instillation in The lack of a completely analogous these animals (G-l14-at ;14). Dr. Hicks treated animals was any different than human experience does not, in itself, characterized this response as "very that for controls. invalidate the Hicks model. It is squally dramatic" (id.). In a subsequent (a) Abbott's Exceptions. The plausible that the increase in tumor experiment, utilizing a more potent dose , exceptions raised by Abbott and the production is not due to the action of of MNU. the MNUtreated animals had a Bureau are discussed below: MNU, but rather Is caused solely by Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61499 cyclamate. The Hicks model may thus studies that are the subject of this there have been no specific allegations be a valid technique for determining the hearing (Tr. at 1(40). The reference to a that the feed may have contained a possible carconogenicity of a test 1.5 mg. dosage of MNUin the second carcinogenic contaminant, Moreover. substance. In light of the "dramatic report appears to refer to the dosage of there was a statistically significant results" found using the Hicks model MNU used in both treated and control difference between the incidence of and-the possible ultimate validation of animals in the two cyclamate MNU tumors in cylamate-treated animals and the Hicks technique for the detection of studies that are the subject of the controls. As is the case with many carcinogens, the increase in tumor cyclamate hearing (Tr. at 1046, 1048J. In factors which complicate carcinogenesis production found in this study raises any event, as Dr. Hicks explained. it is bioassays, the presence of a considerable suspicion as to the safety irrelevant whether a 1.5 or 2 mg. dosage carcinogenic contaminant in the feed of cyclamate. of MNU was used in the cyclamate cannot negate a positive Iinding of Itis possible that the presence of MNU studies that are the subject of the carcinogenesis as long as both the MNU in the bladder of the test animals cyclamate hearing, because both control and treated animals consume the plays a role in the tumor production dosages represent a "noncarcinogenic same feed. Thus. the failure to analyze found in the Hicks study. The Bureau dose". The results obtained from feed does not invalidate the results of argues that that circumstance is not preliminary studies using the Hicks the Hicks study. Itshould be noted. totally unlike certain human model and either of these dosages however, that such contamination can experiences. The Bureau contends that it produced identical "noncarcinogenic" compromise a negative result by causing is reasonable to expect that. ifingested results (Tr. 991-92; 1036-37). Moreover. such a high tumor incidence in the by humans. cyclamate will interact with whenever Dr. Hicks employed this treated animals becomes statistically carcinogens or suspect metabolites in method a control group was utilized insignificant. the bladder (G-114 at 14-15J. MNU is a with the identical amount ofMNU as the (ivJ AllegedFailure to Replicate Dr. nitrosamide which breaks down treated group (Tr. at 992J. SO long as the Hicks' Work:Abbott alleges that Dr. spontaneously to a carcinogen which is difference between treated and control Hicks' model cannot be accepted until it thought to be identical to a metabolite of animals is statistically significant, the has been replicated and that attempts to dimethylnitrosamine [G-114 at 7; G-65). increase can be attributed to cyclamate. do so "have been unavailing" (Abbott's Dimethylnitrosamine in turn can be Thus, Abbott's exception is without Brief at 30).Although it is true that Dr. produced in the urine of people with merit. Hicks' study has not been replicated. the bladder infections (G-114 at 7; G-60; G Abbott further contends that Dr. Hicks attempts of Dr. Mohr to do so were 61). The Bureau further contends that admitted that she did not foUow the incomplete at the time of the hearing [A other carcinogens or suspect metabolites technique described in her publication 842at 2-3). Thus. no final conclusions may also be found in the human bladder (A-804: at 3J. However. Abbott's can be drawn from his work. Until such (G-114 at 14-15). I do not find these contention is based upon a time as valid efforts to replicate Dr. arguments totally convincing. The misinterpretation of a discussion Dr. Hicks' fmdings are unsuccessful. her tumors found in cyclamate treated Hicks had with a Dr. Moore in a round work cannot be dismissed on this basis. animals may have been due in part to table discussion in Geneva. In that Of course, ifDr. Hicks' work is the unique circumstances of this test. discussion. Dr. Hicks was not referring replicated, greater confidence can be Abbott's unsupported argument that to the cyclamate experiments which are placed in her methodology. these unique circumstances are at issue in the hearing, but rather to a Abbott also contends that Dr. Mohr responsible for the tumors found in different experiment in which Dr. Hicks found 2 mg. ofMNU to be carcinogenic cyclamate treated animals is, however. used a batch ofMNU which caused whereas Dr. Hicks found the same insufficient by itself to rebut the tumors (Tr. at 99O-91J. Thus. Abbott has dosage to be ''noncarcinogenic'' suspicion raised by the Hicks study. The failed to establish that Dr. Hicks did not (Abbott's Brief at 3OJ. Abbott claims that Hicks study is sufficiently analaguus to follow the technique described in her this discrepancy "means that something human experience. when considered publication. went wrong with Dr. Hicks' [sic] MNU." with the lack of a convincing (iiiJ Failure to Analyze Feed: Abbott This contention is incorrect. Dr. Mohr explanation negating the strong results contends that Dr. Hicks' failure to used a more active bath ofMNU than found by Dr. Hicks. to cause me to analyze the feed for pesticides and other Dr. Hicks used [Tr, at 99O-91J. As a conclude that the results -ofthe Hicks contaminates is a deficiency in the study result, even though Dr. Mohr used the study cast doubt upon the safety of (Abbott's Brief at 30J. The ALJagreed same dosage ofMNU as Dr. Hicks. his cyclamate. Valid and convincing with Abbott and the Bureau took batch produced tumors whereas Dr. negative direct feeding studies are exception to the ALI's finding. The Site Hicks' batch did not [id.J. In any event. required to rebut this doubt. Visit Report of the Temporary as long as the treatment and control [ii] Criticisms ofthe Hicks Technique: Committee did state that "no analysis of groups receive the same dosage ofl\n-.'U Abbott makes two crtiticisms the feed was made Iorpestlcldcs, from the same batch and there is a concerning the technique employed by mycotoxins, or other contaminants" (G statistically significant difference Dr. Hicks. First, Abbott alleges that Dr. 41, App. m, Hicks. et a1. at 3). However, between the two groups. the data are Hicks was uncertain as to how much the Site Visit Report concluded that "the acceptable (id: Tr. at 587J. MNU she used (Abbott's Brief at 29-30J. facilities, environmental controls. Additional support for the validity of In one report (G-2J. Dr. Hicks refers to experimental design. and conduct of the the Hicks' technique is found in the the usage of a 2 milligram [mg.] dosage study were thought to be adequate to work of other scientists who have of MNU.In a second report (G-64J. use warrant the consideration of the employed methods analogous to those of of a 1.5 mg. dosage of MNU is reported. experimental results" (id. at 10J. Dr. Hicks and obtained favorable A review of these reports and Dr. Hicks' Furthermore, as was the case with the results. Dr. Gilbert Friedell, head of the testimony shows that the reference to a chow diet utilized in the calcium American National Bladder Cancer 2 mg. dosage of MNU in the first report cyclamate portion of the Friedman Program, has employed Dr. Hicks' appears to refer to a pilot experiment study. the same feed was used in both method with a dosage of nitrofuran. which preceded the two cyclamate the treated and the control animals and instead ofMNU, as the initiating 61500 Federal Register I Vol. 45, No. 1-81 I Tuesday, September 16, 1980 I Notices carcinogen [Tr; at 989). Animals MNUwhereas tumors were found in 20% was adequate (G--41, App, III, Middlesex administered this .dosage of nitrofuran . of the rats receiving 1.5 mg. ofa different Hospital Medical School at10). were then fed a saccharin containing preparation ofMNU. This difference in Abbott does not evensuggest why the diet (id.).The incidence of tumors in the tumor production is attributable to the staggered starting times should treated group was approximately 50% differing potency of different invalidate the study. Since equal (id.). Dr. Hicks also testified that ' preparations ofMNU{'fr. at 990-91). numbers of control and treated animals another scientlst.Dr, Bryan, who This alledged inconsistency thus does were started together, even if the MNU appeared as a witness for the Bureau, not render the experiment invalid, was unstable and broke down during conducted an experiment in which MNU provided, as was the case inboth of Dr. the course of the experiment, it would was used to initiate a carcinogenic Hicks' experiments with cyclamate, that not affect the validity of the study. Such response in the epithelium [Tr, at ~89). both treatment and control groups in a study would evaluate the carcinogenic Using this method, Dr. Bryan each experiment receive the same response of cyclamate under varying demonstrated a synergistic effect dosage ofMNU from the same potencies of MNU. The difference between a tryptophan derivative and preparation, and that there is between treated and control anlmals MNU (id.). (Abbott attacks this part of statistically significant difference would still be attributable to cyclamate, Dr. Hicks' testimony on the ground that between the. treatment and control More importantly, there is no reason to Dr. Bryan never mentioned this groups. suspect that the MNU used by Dr. Hicks experiment in his written or oral did break down during the course of the testimony (Abbott's Brief at 30,n.t), In support of this argument, Abbott attempted to present, by way of rebuttal, experiment, Dr. Hicks testified that However, Dr. Bryan was never asked precautions were taken to ensure tho about this experiment, nor did Abbott the testimony of Dr. Deutsch-Wenzel. Dr. Deutsch-Wenzel would have stability of the MNUpreparation used in seek to have this question posed to Dr. the cyclamate experiments (Tr. at 992). Bryan following Dr. Hicks' testimony.) testified that the "potency" ofMNU does notvary from preparation to When a bulk batch ofMNU was The Bureau also notes that Dr. Hicks received by Dr. Hicks, it was achieved negative results with the . preparation. Although this testimony was properly excluded by the AL], I immediately weighed into small allquots known noncarcinogens coffee and (id.). Each one was sealed in a glass cyclophosphamide {Bureau's Brief at 33). have nevertheless considered it. Even assuming, arguendo, that the potency of bottle, wrapped in foil to keep out light Although the experiments involving (1) and stored at minus 20 degrees (id.). nitrofuran (rather than MNU) and a 'MNU does not vary, itis irrelevant to the validity of the Hick's experiment so This procedure maintained tho stability tryptophan derivative and (2) MNU and of each batch (Tr. at 993). The procedure , coffee or cyclophosphamide (rather than long as both the treatment and control groups received the same preparation'of followed for dosing tho treated and . cyclamate) cannot be consideredirue control animals was the same in every replications of Dr. Hicks' experiment, MNU and it was handled the same way. Ifthe procedure is followed, the case (Tr. at 1071;1073). Thus, there is no they lend some support to the validity of reason to believe that the staggered her method. difference in tumor incidence between In sum, I find that there are no treated and control groups can be starting times of animals entered in tho study would diminish the reliability of unsuccessfulattempts to replicate Dr. attributed to cyclamate. Dr. Hicks the study's results. Hick's experiment In view of the fact repeatedly testified that the same MNU that Dr. Hick's technique may yet be was given to treatment and control (viii) Alleged Differences in Numbers validated, and the fact that the, groups and it was handled the same ofAnimals Examined at Histology: Ono technique has heen successfully used way (Tr. at 991-93; 1020;1030;1070-72). report of Dr. Hicks' cyclamate with other carclnogens and Thus, the alleged constant potency of experiment (G-2) states that 54 animals noncarclnogens, I do not 'consider the MNUis irrelevant. were treated with MNU and cyclamate Iack of a successful replication a ' whereas a second report of the study (vii) Failure to Start AllAnimals at (A-832) states that 69 animals were significant deficiency. ~eSameTIme~d~eMk~d (v) AllegedLack ofFormal treated with MNUand cyclamate. The Uncertainty as to the Number of discrepancy is attributable to the fact Randomization: Abbott contends that Animals Started at Various Times Dr. Hicks failed to randomize the that the first report was a preliminary During the Test: Abbott contends that report, whereas the ~econd report, animals in J.1er study (Abbott's . these factors are significant in. Exceptions at 23). Dr. Hicks explained, which lists a larger number of animals, evaluating the validity of the Hicks' however. that although she did not use a is the fmal report. 'f.he discrepancy .experimental procedure (Abbott's. table of randomization, she did use a appears to be due to the fuct that all system ofrandomization (Tr. at 1072). Exceptions at 23-24). animals had not been sorted or had not The Site Visit Report found that "It is The animals in the Hicks' study were yet been examined at the time of tho unlikely, that any biases were introduced entered into the study over a period of preliminary report (Tr. at 1064). by the lack of a formal randomization months (Tr. at 1053). Whenever (ix) Slide Examination: Abbott method being used to assign the animals vacancies for storing the animals questions whether the examination of to the experimental groups" (G--41, App. became available, a palred group of the slides to verify the existence of III, Hicks, et al. at 8). As noted above, control and treated animals were added tumors was totally blind, l.e; whether the Temporary Committee concluded to the study (Tr. at1052-53). Although the investigator could determine that the that the experimental design and Dr. Hicks could not recall the number of slide came from a treated or control conduct of the study were adequate (id animals started at various times during animal (Abbott's Exceptions at 23). Dr. at 10). I agree with the-conclusions of the study, she testified that her records r Hicks explained that although the slides the Temporary Committee. would reflect when each animal was examined for purposes of histology were (vi) Alleged Inconsistency in Tumor entered into the study [Tr, at 1054-55). numbered, they were not numbered in Findings: Abbottquestions the validity Many of these records were examined the same way as the animals were [Tr; of Dr. Hicks' technique on the ground by the Temporary Committee's Site Visit at 1066). There was no way of that no tumors were found in rats Team (Tr, at 1055).The Site Visit team determining from the number on tho receiving 2 mg. of one preparation of concluded that the conduct of the study slide whether the animal came from the Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61501 control or treated group (id.). Thus: for correctly found that Abbott could and Abbott's contention that there was a all practical purposes. Dr. Hicks' should present the matters it sought to discrepancy in the amount ofMNU used examination of slides was blind. Even if introduce as rebuttal testimony 8S by Dr. Hicks. This contention is based there is some question about the argument in its brief (Order dated on two publications (G-2, G-64) that histology employed by Dr. Hicks. that September 12, 1977).Abbott did present were available to Abbott prior to the histology was confirmed by the Site each of these arguments in its hearing: (Ii)Abbott's allegation that the Visit Team of the Temporary Committee exceptions (Abbott's Exceptions at 23) Hiclcs' study lacked formal (G-41. App. ill, Middlesex Hospital and they have been considered above. randomization. This issue was Medical School at10). The Temporary Thus, Abbott has not been prejudiced by mentioned in the Temporary Committee Site Visit Team concluded the exclusion of its rebuttal testimony, Committee's Report and therefore was that the histopathologic examinations Moreover, as discussed above, that known to Abbott prior to the hearing: were satisfactory (id. at 10). Thus, rebuttal testimony was properly (iii) the alleged discrepancy in the Abbott's criticism is groundless. excluded. number of bladders each of the two (x) Cyclophosphamide: Dr. Hicks (i) Testimony ofDr. Deutsch-Wenzel, reports states were e.xamined (G-2. G conducted an experiment in which Dr. Wenzel was to testify concerning the 64).This information was also known to treated animals received MNU plus properties ofMNU, namely that its Abbott prior to its cross-examination of cyclophosphamide and control animals potency does not vary from batch to Dr. Hiclcs: and (iv) the fact that Dr. received MNU.The cyclophosphamide batch and as to its volatility. This Hicks used her technique with MNU did not produce any tumors in the evidence is irrelevant. As discussed cyclophosphamide and considered treated animals (G-2 at 225).Dr. Hicks above, the "potency" and "volatility" of cyclophosphamide a noncarcinogen. cited this evidence as support for the MNU are irrelevant so long as both This information is contained in a report validity of her-technique on the ground treatment and control groups receive of a study (G-2) which was filed y,ith that "there is no evidence to establish MNU from the same batch and that FDA's Hearing Clerk prior to the that cyclophosphamide is a bladder MNU is handled the same way. hearing. Abbott thus had ample carcinogen in man" (Th.at 1067)or rats (Ii) Testimony ofDr. Mohr: Dr. Mohr's opportunity to submit testimony on (Tr. at 1068). Two other Bureau rebuttal testimony concerning his these issues prior to the hearing and has witnesses also confirmed Dr. Hicks' attempts to replicate Dr. Hicks' works no grounds to claim that that testimony opinion that cyclophosphamide is not a and how his results allegedly conflicted was proper rebuttal. bladder carcinogen in rats [Tr, at 555i with those of Dr. Hicks were admitted Abbott also sought to introduce 965-66).Abbott contends that Dr. Hicks . into evidence. The ALJonly excluded rebuttal on statements by Dr. Hicks that is incorrect about the Dr. Mohr's general criticisms of the slides were examined blind "to a large noncarcinogenicity of Hicks' model. This testimony was extent" (Tr. at 1006)ithat the "basis of cyclophosphamide and the the ALJ properly excluded because Abbott did the model is to initiate neoplastic improperly excluded the testimony of not properly state its intention ahead of time to introduce this testimony in changes with MNU" (G-114 at 8)i that Dr. Schmaehl on this issue. For the "in focal areas the epithelium is reasons discussed immediately below in rebuttal (See Motion For Leave to Adduce Rebuttal Testimony; Docket No. stripped" (Tr. at1067)and that all Subsection (b), I find that Dr. 117, p. 5). Since these general criticisms animals in the test were not started at Schmaehl's testimony was properly . the same time (Tr. at 1051-52). These excluded. I therefore find that Dr. Hicks' went beyond the scope of Abbott's request to adduce rebuttal testimony, statements, standing alone, do not cyclophosphamide experiment was ~ entitle Abbott to rebuttal. Abbott failed properly considered by the ALJand the ALJproperly excluded pp. (beginning with 11, p. 4 through the last to provide the ALJwith any explanation, does lend support to the validity of the sentence on p. 6) of Dr. Mohr's let alone a comincing explanation, of Hicks' technique. testimony (Order. Docket No. 149, why these factors are so important as to Moreover, even ifDr. Hicks' November 21,1977).Moreover, even if require expending the addtional time cyclophosphamide experiment is Abbott had properly requested leave to and resources to hear additional excluded from consideration because of adduce this rebuttal testimony, these rebuttal testimony. Abbott did not the possibility of that substance's matters could have and should have provide any information as to how these carcinogenicity, Dr. Hicks also been introduced as part of Abbott's factors affect the validity of Dr. Hicks' performed a study using her technique written direct testimony. technique. In the absence ofsuch an with coffee and obtained negative (iii) Testimony ofDr. Schmaeh/: Many offer of proof, Abbott has failed to results [Tr, at 553).Thus, even without of the alleged deficiencies which Abbott provide sufficient justificationfor considering the results of the sought to have Dr. Schmaehlteslify rebuttal. To the extent that these cyclophosphamide experiment, there is about were known to Abbott prior to the statements by Dr. Hicks, on their face. a study providing a negative correlation cross-examination of Dr. Hicks. The indicate that her technique is invalid, for Dr. Hicks' technique and supporting procedural regulations governing the Abbott does not need rebuttal witnesses Dr. Hicks' conclusion that tumors found cyclamate hearing required the Bureau to restate the obvious. These matters in animals receiving MNU followed by to file with FDA's Hearing Clerk, at the have been considered above and do not cyclamate should not be attributed time of publication of the notice of raise any meritorious questions as to the solely to MNU. hearing, all documentary data and validity of Dr. Hicks' technique. (b) Abbott's Request for Rebuttal. information upon which it relied. 21 CFR (c) Bureau's Exceptions. The ALJ Abbott claims that it was prejudiced 12.85(a)(3). questioned the validity of the Hicks' because it was prohibited from The majority of issues for which study on the ground that no separate introducing rebuttal testimony from Drs. Abbott sought to introduce rebuttal control groups were anesthetized 01" Schmaehl and Deutsch-Wenzel, whom were contained in two reports (G-2, G instilled with an innocuous material. - Abbott asserts would have testified as 64) which were filed with FDA's Hearing The Bureau talces exception to this to the alleged deficiencies in the Hicks' Clerk pursuant to 21 CFR12.85{a)(3) • 'flndlng, The Bureau contends that there model. In denying Abbott's motion to prior to the hearing, or were otherwise is no evidence in the record that adduce this rebuttal testimony, the ALJ known to Abbott. These issues are: (i) anesthetization or instillation could 61502 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices affect.tumor production. The Bureau or was noncontributory" (G-41 at 21). effects seen in the Oser study, the . further contends that the use of a control The Temporary Committee further found Bureau's arguments do not eliminate the group that was anesthetized and that "[o]f particular concern is the [Dser possibility that saccharin or possibly . received MNU is sufficient because v study] in which a statistically significant CHA contributed to or was the sole there is no evidence that the treated and increase in bladder-tumors occurred in cause of the carcinogenic effects seen In control animals were handled any animals treated with a mixture of the Dser study. The fact that studies of differently (Bureau's Exceptions at 3). cyclamate and saccharin" (G-41 at 48 saccharin have not shown an effoct at Abbott contend~ that these variables 49). the feeding level of saccharin utilized in "can disrupt the orderly interpretation of The ALJfound that "the study might the Oser study (1%) does not eliminate a study's results, and perhaps make a 'be relevant to showing cyclamate's the possibility that saccharin played a study's results impossible to interpret cocarcinogenic potential * * *" (IDat role in the carcinogenic effects found. meaningfully" (Abbott's Reply at 4-5). 18).The ALJfurther found that "there Thresholds for carcinogens have not Abbott further contends that the are many confounding variables been established (see Tr. at 1066-69). manner Iri which the animals were (presence of saccharin, CHA, bladder Moreover, it is unknown whether the anesthetized and subjected to MNU was parasites and calculi and the design) species, strains and conditions of the never adequately explained by Dr. ' that prohibit relying on the study to saccharin studies on which the Bureau Hicks (id.). This latter point is different show the carcinogenicity of cyclamate" relies are the same as the species, from that of the ALJ. The ALJdid not (id.). find that the method of anesthetization strainsand conditions of the Osor study•. 3.Analysis: In its exceptions, Abbott The possible effects of saccharin In tho or instillation of the MNU was different contends that the uncontrolled variables for controls or treated animals. Indeed, Oser study therefore cannot be in the Oser study make it inappropriate eliminated. The same, of course, holds Dr. Hicks testified that "the controls for assessing either the carcinogenicity were given exactly the same treatment true for cyclamate, particularly In view or cocarcinogenicity of cyclamate. of Hicks and Bryan studies and with MNU prepared in exactly the same Specifically, Abbott contends that (1) way, from the same batch, at the same recurrent findings of bladder tumors the study was not intended to examine found in the direct feeding studies pH, in the same medium, very often from carcinogenicity; (2) the prese:qce of the same individual solution [as the (Friedman, Schmaehl, Homberger and saccharin and CHA are uncontrolled _ Hicks (direct feeding) studies). All of treated group]" (Tr. at 1020). Thus, variables; (3) there was no data on Abbott's latter point is without merit. these findings add credence to the ~ possible trace impurities.in either the The basis for the ALI's criticism is possibility that cyclamate was the sale apparently the statement in the saccharin, the cyclamate, or the or primary cause of the production of Temporary Committee's Site Visit cyclamate/saccharin mixture; (4) the bladder tumors in the Dser study. The Report that "[n]o anesthetized control presence of bladder calculi in many of design of the Oser study is certainly not . the rats with tumors complicated any groups or ones instilled With an ideal for determination of the innocuous material were 'established" finding of a direct causal connection between the tumors and the test carcinogenicity of cyclamate, but it does (G-41, App, III, Middlesex Hospital raise a suspicion as to cyclamate's Medical School at 3).The Site Visit substance; (5) spontaneous tumors couId have been caused by bladder parasites; safety and requires a close exumlnatlon Report concluded, however, that "[t]he of the studies involving the direct facilities, environmental controls, (6) the pathologists differed in their dlagnosls of the tumors, agreeing feeding of cyclamate, experimental design, and conduct of the The suspicion raised by the Dser study were all thought to be adequate to unanimously on only 4 of the 12 rats . diagnosed; and (7) attempts by study could be rebutted by valid and warrant consideration of the convincing negative studies. I do not experimental results" (id at 10). N9 Schmaehl, Ikeda and Kroes to replicate expert testified that these factors cast the Oser study have been unsuccessful find persuasive, however, Abbott's doubt onthe validity of the Hicks (Abbott's Exceptions at 19, 20-22). arguments that studies by Bchmaehl, experiment. More importantly, as noted The fact that the Oser study was not Ikeda and Kroes represent such studies. above, the procedure for instilling the , specifically designed to determine the First, as previously noted, at a minimum, MNU in treated and control animals carcinogenicity of cyclamate is to disprove results that are inconclusive was the same. Thus, the anesthetization irrelevant except insofar as specific but suggestive of a positive effect, tho and instillation of MNU.were controlled aspects of the design or conduct of the test substance must be tested in the variables and these factors do not cast study can be shown to make it more same strain of the same species as used doubt on the validity of the study. difficult or impossible to draw in the experiment with positive results. c. Oser, et al. (G-81). (1) Study Design: conclusions concerning the The Oser study involved Wistar dorlvcd This study involved Wistar-derived carcinogenicity of cyclamate from the FDRLrats whereas the Kroes study FDRL strain rats in groups of 35 males study, Abbott alleges that it has found involved mice and the Bohmaehl study and 45 females fed a diet containing a two such defects. First, Abbott contends involved Sprague-Dawley rats. mixture of ten parts sodium.cyclamate that the presence of saccharin and CHA Moreover, the Kroes study was posltlvo to one part sodium saccharin. After 78 complicate any findings with respect to for lymphosarcomas and the Schmaehl weeks on study the animals were 'cyclamate. The Bureau contends that it study cannot be considered negative subdivided and some were additionally is highly unlikely that effects seen are because there was one tumor found in treated with cyclohexylamine due solely to saccharin because (1) the the cyclamate treated group in that hydrochloride. ' ratio of cyclamate to saccharin in the study (see Section IV.B.2.a.(5J). Although (2)Study Results: Bladder tumors mixture used in the Oser study is ten to the lkeda study involved the same strain were reported in 12 of the 80 rats given one and (2)saccharin has not shown a and species of rats as the Oser study, the high dose of the treatment mixture. carcinogenic effect at the feeding level the study is inconclusive because of the The Temporary Committee found that utilized in the Oser study Bureau's Brief low sensitivity of the study and the fact "no conclusion can be made 'as to at 35; G-120 at 14). Although it may be that histopathology had not been whether cyclamate was the causative probable that cyclamate was the sole or performed on all organs (see Section agent, acted in concert with saccharin, primary cause of the carcinogenic IV.B.2.a.(2)). Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61503 I also disagree with Abbott's probable cause of carcinogenic effects in parasites" and that although "parasites argument concerning the alleged the study and that therefore. the study were found in every animal" .. .. no impurities in the cyclamate and does raise a suspicion as to the safety of bladder tumors were found" (ID at16). saccharin mixture used in the Oser cyclamate. c. Analvsis: Abbott contends that the study. I find persuasive the Bureau's C. Negative Studies "a:nsurvival figure" is incorrect: that it contention that there is no evidence that is nowhere mentioned in the published such impurities are present or what 1. Gaunt, et al. (:1-706). a. Study study and that it cannot be drawn from significance they might have (Bureau's Design: This study involved groups of the tables on the report of the study Brief at 37). Dr. Oser states that "The SPF Wistar rats, with 48 of each sex in a (Abbott's Exceptions at 30). I agree 'with possibility of the presence of impurities group, fed diets containing eIther 600, Abbott that the 32.~ survival figure is and their effect cannot be overlooked in 2,000 or 6,000 ppm of cyclohexylamine incorrect. The correct survival rate is, light of later developments particularly hydrochloride. The stuy was conducted however. nevertheless unusually low. with respect to commercially produced for 104 weeks. All major organs The authors of the Carson study made saccharin" (A-803 at 4-5). Dr. Oser does including bladders were microscopically the following comment on the survival not state the identity of these suspected examined. rates of the animals tested: impurities or whether they would be b. Study Results: No tumors were expected to have a carcinogenic effect found. The authors of the study Survival ranged from ao-gG':3 up 10 the 7Bfu week in all groups except the control females Moreover, Dr. Oser does not explain his concluded that "[t]here was no indication of a carcinogenic effect at any where it was 71'b. Toward the end of the reference to "later developments" secondyear, mortality increased. the concerning "commercially produced of the levels of treatment" (A-611 at 2). terminal survival rates for all lest groups saccharin," nor does Dr. Oser state The Temporary Committee found that averaging 45.1 and 55.9%for the males and whether or not the saccharin used by "the test material did not display females. respectively (v.ith no grading relaled FDRL was "commercially produced." carcinogenicity" (G-41 at 17). to dose level). compared with 46:0 sunival The cyclamate/saccharin mixture used c. Analysis: Abbott and the Bureau for each sex in the controls. in the study was supplied by Abbott (G agree that the resuts of the study were negative in terms of carcinogenicity (G-4 at 50.) Thus. although the ALrs 81 at 4). Surely, if there was some "32% survival figure" is incorrect. the reason to suspect the presence of (Abbott's Brief at 19; Bureau's Brief at 19). ALJ is nevertheless correct in that a impurities in the cyclamate of saccharin, large percentage, approximately 50%. of Abbott would be in a position to provide I find that under the conditions of this test, cyclohexylamine did not display the animals in the study did not survive more specific information as to the the first two years of the study. These nature of those impurities. I find that the any carcinogenicity. I note, however, that although studies of the average 45.1 to 55.9% survival figures information provided by Dr. Oser is not represent a reduction in the sensitivity sufficiently specific to justify cyclohexylamine metabolite of cyclamate are relevant to the safety of of the study because the animal!> whi.ch questioning the validity of the study. died prior to the termination of the study The record does not support Abbott's cyclamate, such studies, standing alone, might have developed tumors had they contention that bladder parasites or are insufficient to establish the safety of survived. (Only the animals that _ bladder calculi were responsible for the other metabolites or of cyclamate itself. tumors found in the treated animals. Moreover,1he stuy is inadequate to survived were evaluated.] Even ifall the Only three of the tumors in the treated rebut questions raised by other studies animals had survived and had been group were associated with calculi in other species or strains of animals evaluated, the study had only a 5O'J' (G-41, App. VIT, Oser et a1.at 3; G-12O (see Section IV.B.l.a.). (I also note that, I chance of detecting, at the 95% at15). This factor supports the agree with the Bureau that pulmonary confidence level, a true difference in conclusion that calculi are not necessary effects and testicular atrophy were tumor incidence of approximat2ly 9% for tumor development (G-12O at 15). In strongly associated with dose level in between the controls and the high dose addition, a dose response relationship this study.) (150 mg/Ieg) treated animals [G--41 App. between the cyclamate/saccharin 2. Carson, et 01. {G-4]. a. Study V, at 20).I find that this low sensitivity mixture and the incidence of tumors was Design: This study involved Ilve groups significantly reduces the confidence that found (G-114 at 28). This does response of 30weanling FDRL Wistar rats of each can be placed on the results ofthis relationship would tend to negate the sex 0, 15, 50, 100 or 150 mg./kg. study. Moreover, although studies of the likelihood that calculi or parasites cyclohexylamine per day. The study cyclohexylamine metabolite of caused the tumors found in the Oser was conducted for 113 weeks, after cyclamate are relevant to the safetyGf study because calculi or parasites would which the survivors were sacrificed. cyclamate, as noted above, such studies. be expected to cause tumors in all Most of the bladders were examined standing alone. are insufficient to groups with approximately the same microscopically. establish the safety of cyclamate. frequency (id). b. Study Results: No tumors were With respect to the presence of Finally, I do not find persuasive found. The authors of the study stated bladder parasites, Abbott ctmteOOs that Abbott's criticism of the Oser study on that tIle]xamlnatlon ofmultiple sections "if a study is to have relevance on the ground that there was a lack of (about 16 to 20) of the urinary bladder whether parasites cause bladder tumors. unanimity on all tumor diagnoses. The from each rat revealed no evidence of the length of exposure to parasites must NCI Pathology Working Group tumorigenesis" .. .. It is of particular be known" (Abbott's Exceptions at 31). confirmed the diagnosis of all twelve interest to note the absence of any Abbott concludes that the Carson study tumors (G-41, App. III), all of which bladder carcinoma despite the intensive is "meaningless on this issue" because were originally reported by FDRL. I examinations that were carried out" . the length ofexposure to parasites is therefore see no reason to question the (A-274 at 28). unknown. I agree with Abbott's diagnoses of tumors in the Oser study. The ALJ noted that only 32%of the exception. It is important to note. . In sum, I find that the validity of the animals survived the two years" of the however, that the length of exposure to Oser study was comprised somewhat by study (ID at 16). The ALJ also stated that parasites is also a factor to consider in the presence of saccharin and CHA. I "[0] ccasionally the incidence of tumors those studies where Abbott argues that find, however, that cyclamate is a is related to the presence of bladder the occurrence of tumors may be due to 61504 Federal Register I Vql. ,45, No. 181 I Tuesday, September 16, 1980 I Notices bladder parasites or bladder calculi drinking water as follows: to seven females [G-41, App, III,Site Visit Report for the rather than the test substance. for four weeks before mating, to five females Eppley Institute for Research in Cancer between the time of mating and delivery, to and Allied Sciences at 4-5: emphasls D. Deficient Studies five females after-mating and continued for added). The following studies do not contain 25 days after delivery, and to four females for I find iliat the above quoted language four weeks before mating and and continued results from which responsible until delivery. from the Site Visitor's Report fully conclusions as to the safety of The study was continued for the lifetime of supports the ALI's statement. Moreover. cyclamate can be drawn because of the F-1 generation, consisting of13-~5 I find that the questionable sensitivity of 4,!eficiencies in the design or conduct of hamsters per group. The study had not been the hamster to detect relatively weak the studies. completed at the time the NCI Temporary bladder carcinogens should be 1. Altof!study (A-691). a. Study Committee made their report. considered and further limits thevalue Design: This study involved levels of h:StudyResults: After 80 weeks no of both Altoff studies. .156,.312,.625and 1.25% of sodium tumors were found in the hamsters that d. Othermatters: As part of the cyclamate or calcium cyclamate given in had died [ill at 14-15). The Temporary Remand Order in this proceeding, the drinking water to groups of 30 male and Committee found that the study "has parties were asked to submit any datu 30 female hamsters. The study was limited value in that none of the animals pertaining to this study. That was done continued for the lifetime of the were continued on treatment for'thelr on September 17. 1979.A review of the • hamsters. lifetime" [G-41 at 19). The ALJ found data did not, however. reveal any b. Study Results: The authors of the that this study "has'limited value with significant effects. study stated that "[t)he present . respect to carcinogenicity because of the 3. Colston, et al. (A-207l. a.Study experiment in Syrian golden hamsters low-dose level and that no F-1 group Design: This study involved a group of adds to the volume of negative evidence was treated beyond 25 days following rhesus monkeys fed an oral dose of on carcinogenicity of saccharin and birth" [ill at 15). . sodium cyclamate, 6 days per week 'cyclamate" [A-691 at 23). c. Analysis: Abbott argues with the since January 1968.As of June 1975,only The ALJ found that the study was ALI's assessment that the second Altoff three monkeys were still being studied. negative [ill at 31),but noted that "[t)he study.was incomplete at the time the b. Study Results: The Temporary e sensitivity of the study was limited by Temporary Committee wrote its report Committee found that "[t)he value of the small group size and the poor [Abbott's Exceptions at 28). Abbott also this study to assess the carcinogenicity survival rate. Less than 15%of the agrees with the Temporary Committee's of cyclamate is severly limited as a animals were alive after 74 weeks" [ill result of the small number of anlmuls at 14). . finding that the second Altoff study has "limited value" [id. at 28-29). Abbott used, the low dose level tested, and tho c. Analysis: Abbott contends that the . relatively short portion of the monkey a' study is negative, relying on the further states that, contrary to a statement in the Initial Decision [po 15), life span studied" [G-4 at 16). conclusion of the authors of the study c. Analysis: Abbott designates the [Abbott's Brief at 18). In its exceptions, Abbott does not contend that the second Colston study as being negative, Abbott contends that the NCI Site Visit Altoff study reinforces other negative (Abbott's Brief at 18). The Bureau agrees group found that the "small initial group fmdings [id.).The Bureau states that the with the findings of the Temporary size limited somewhat the sensitivity of study is "worthless" because it was not Committee [Bureau's Brief at 13; G-126 the study A-647 App.Ill at 4", and that a lifetime feeding study [Bureau's Brief at 9-10). For the reasons stated by the therefore the ALJ distorted the evidence at 15). Temporary Committee, I find that this and is prejudiced because he found that 'Abbott takes exception to the ALI's study does not contain results from the sensitivity of the study was statement that the "NCI Temporary which responsible conclusions as to tho "limited" [Abbott's Exceptions at 27-28). Committee Site Visitors suggested that safety of cyclamate can be drawn. The Bureau notes that the Temporary hamsters were not sensitive enough to 4. Fitzhugh (A-19Z). a. Study Design: Committee found that "[t)he small detect weak carcinogens" [ill at 15). This study involved Osborne-Mendel number of effective animals, resulting . Abbott contends that this statement is rats, in groups of ten males and ten • from high early mortality ofthe treated not synonomous with that of the females, fed 0, 0.01%, 0.1%, 0.5%, 1.0%or hamsters, reduced the sensitivity of the Temporary Committee and therefore 5.0%sodium cyclamate, The study was , study" [Bureau's Brief at 14). The Bureau prejudicial to Abbott [Abbott's conducted for 24 months. The bladders concludes that the study is inconclusive Exceptions at 28). This contention is were not microscopically examined, [id. at 19). , groundless. The Site Visitors stated that: except for those animals in the high I disagree with Abbott's exception. The hamster has proven to be a good dosage group. The terms "limited" and "somewhat animal model to demonstrate the b. Study Results: The Temporary limited" are roughly synonomous, In any carcinogenicity of some bladder carcinogens. Committee considered this study event, the record supports the ALI's When beta-napthylamine (a relatively strong "deficient is that the bladders were statement. This Altoff study had only a carclnogenjwas initially tested in the examined only microscopically" [G-41 50% chance of detecting, at the 95% hamster, no'carcinogenic response was elicited. although the dose was rather large at 20). confidencelevel, a true difference in (0.1%). It was found that even a higher dose The ALJ found that the failure to tumor incidence of approximately 38% (1.0%) was needed to produce a carcinogenic microscopically examine all bladders • between the controls and the high dose response. In viewofthese results, itmustbe "seriously questions the validity of the , treated animals [G-41, App, V, Table IV questioned whether the hamster also is a negative results" (ill at 12). at 19). This study is therefore too goodanimalmodelfor detecting relatively G. Analysis: Abbott did not take insensitive to be considered proof of weak bladdercarcinogens. . exception to the ALfs finding. The safety. The questionable sensitivity ofthe hamster Bureau contends that the Fitzhugh study 2. Altof!et 01. ("Second Altof!study") to detect relatively weak bladder carcinogens, the ratherP90r sensitivity ofthe is deficient [Bureau's Brief at 15, 19). (G-41 at 19). a. Study Design: The ALJ studies referring to both Altoffstudies) and The lack of microscopic examination described this study as follows: the low dose levels testedmust all be of bladders in the Fitzhugh study raises ••• • Syrian golden hamsters were given consideredin determining the value ofthe the possibility that tumors may have 1.5%sodium or calcium cyclamate in their results obtained. been overlooked. Therefore. I cannot Federal Register I Vol. 45. No. 181 I Tuesday. September 16. 1980 I Notices 61505 draw any valid conclusions concerning conduct and results are known" (G-41 at in groups of50 females each. One group the safety of cyclamate from this study. 19). was exposed to 5%sodium cyclamate It should be noted that my decision to C. Analysis: In view of the lack of pretreated with polyethylene glycol. The place no reliance on the results of this information on this study. it is of no use othergroup received 5%sodium study is not inconsistent with my finding in determining the carcinogenicity of cyclamate pretreated with polyethylene that the Rudali study. in which no cyclamate. glycol and benzo(aJpyrene. The mice histopathology was performed, is d. OtherMatters: As part of the were not randomly allocated in that the inconclusive but suggestive of a positve Remand Order. the parties were asked older mice were placed in the control effect. In the Rudali study. tumors were to provide a report of this study. The group. visible macroscopically. As I explained only information provided was a review b. StudyResults: The Temporary in Section IV.B.l.a.• even if microscopic article which discussed the early studies Committee considered this study tumors were present in control animals on artificial sweeteners. deficient because bladders were not in the Rudali study. the large lung and 7. Adamson (G-41 at25). a. Study examined microscopically (G-41 at18). liver lesions visible in cyclamate treated Design: This study involved twenty c. Analysis: Abbott initially listed the animals without histopathology are an treated monkeys (three groups of 5 of Roe study as negative. but did not indication of a more rapid time of onset each sex) fed O. 100 or 500mg/kg/day of 'discuss it in its brief to the ALJ (Abbott's of the tumors and thus are supportive of sodium cyclamate. The study had been Brief at18). a finding of carcinogenicity. In contrast. ongoing for five years at the time of the The Bureau listedthe Roe study as in the Fitzhugh study. tumors may have Temporary Committee Report. "inconclusive" (Bureau's Briefat10). been present in cyclamate treated b. Study Results: The Temporary One Bureau witness. Dr. Cranmer, animals but overlooked due to lack of Committee found that "no conclusions stated that the lack of microscopic histopathology. Unlike the Rudali study. regarding cyclamate's potential • examlnatlon of animal bladders was a where the presency of microscopic carcinogenicity in monkeys can be made "major flaw" in the Roe study (G-126at tumors in control animals would not until either a response is detected or the 10J•• substantially alter the interpretation of study is terminated" (G-41 at 25). The In the Remand Order. the parties were the results. a finding of microscopic ALI reached the same conclusion (ID at asked to comment on the tumors in cyclamate treated animals 15). maldistribution ofweight or age in the and none in controls could result in a c. Analysis: Anbolt did not take various groups in the study. In response, finding that the Fitzhugh study was exception to this finding. I agree with the parties stipulated that "since positive. This possibility prevent me the Temporary Committee's findings and complete histologic examination was from concluding that the Fitzhugh study conclude that valid conclusions as to the not conducted. this study does not is negative. safety of cyclamate cannot be drawn contribute to an assessment of the from this study. carcinogenic potential, ifany, of d. OtherMatters: As part of the 8. Industrial Biotest (A-394-4fXJ). a. cyclamate" (Remand Proceedings Remand Order. the parties were asked Study Design: This study involved to submit tpe data for the Fitzhugh Stipulation of the Parties at 4). I agree Beagle dogs in groups of each sex given 'with the parties. and. consequently. study. A review of that data did not cyclohexylamine sulphate (CHA-S) as a have not considered this study in my reveal any significant effects and 25%mixture with lactose for the first 193 assessment of the safety ofcyclamate. confirmed that all bladders were not weeks. and as a 50% mixture with microscopioally examined, lactose for weeks 194-400. The first Eo OtherBvidence 5. Schmaehl et ale (A-386A). a. Study group was fed a concentration of 0.15 In addition to the animal studies Design: This study involved groups of mg CHA-8/kg/lweek which was discussed above. the record contains Sprague-Dawley rats fed butylbutanol increased to 50 mg/kg/wk at the 194th studies performed in test tubes (liz lri/rO) nitrosamine (BBN) or BBNplus sodium week. For the second group. the and retrospective studies on the use of cyclamate. concentrations were 1.5 mg/kg/wk and cyclamates by humans (epidemiological b. Study Results: There was a l00'Jb 100 mg/kg/wk. and for the third group studies). The ALJ found that the in vitro incidence of bladder tumors in both 15 mg/kg/wk. tests "represent no more than a treatment groups (G-41 at 23).The b. Study Results: The study is predictive tool. and in light of the results Temporary Committee concluded that incomplete in that seven of the original of the animal feeding studies. cannot be the study "has limited value in providing sixteen animals were still being tested considered as determinative of the issue evidence for or against cyclamate's at the time of the Temporary Committee , of carcinogenicity" (ID. at 33J.As to the potential carcinogenicity" (id.). Report. The Temporary Committee epidemiological studies. the ALI found C. Analysis: Because of the 100% found that "[u]nless a statistically that the studies inquired about artificial incidence of tumors in both groups, I significant carcinogenic response is swcetners and did not distinguish find that the study is deficient. demonstrated. this study has limited between saccharin and cyclamateid.). 6. Bar (A-131). a. Study Design: The value in assessing the carclnogenlclty of The ALJ further found that the senshity ALI states that "[t]his study employed cyclohexylamine due to the small of these studies was severely limited rats thatwere laboratory-bred and were number of test animals" (G-41 at 24). because at the time the studies were "fed these [sic] doses of sodium c. Analysis: Abbott agrees that the conducted cyclamate had only been on cyclamate;150 mg/kg/day, 300 mg/kg/ Industrial Biotest study was incomplete the market for five years and subjects aay and 450 mg/kg/day in groups of at the time the NCI Committee wrote its were generally questioned within five males and females ranging from 30-55" report. I fmd that this study does not years. Because the proposed use of (ID at12). contribute to the evaluation of the safety cyclamate would result inlifetime b. StudY,Results: The Temporary of cyclamate because the study is not exposure and because carcinogenic Committee stated that "[v]ery few completely reported and the number of effects often do not manifest themselves details on this study were available to animals involved is too small for the for periods of 25 to 30 years after the Committee. Thus, an evaluation of it study to be of any value. ex..posure, the ALI concluded that "no' cannot be made until after it is . 9. Roe. et al, (A-286). a. Study Design: conclusion concerning cyclamate's completed and details of its design. This study involved Swiss albino mice safety can be drawn on the basis of 61506 Federal Register I Vol. 45, No. 181 I Tuesday, September ~6, -1980 / Notices these studies" (ID at 33).Abbott and the Dr. Epstein. another Bureau witness, ' The repeated nature of these findings Bureau did not take exception to the stated: across such a variety of species, colla, ALI's conclusion with respect to the in. EPAtakes the position tliat-mutagenesls.is and laboratories greatly enhances thoir vitro and epidemiological studies. I an extremely critical public liealth risk. credibility (see Subsection D below). concur with.the conclusions, ofthe ALJ because its effects may extend to a large Although these chromosome uberratlons with respect to these studies. number of generations to come. were predominantly "breaks" which do. not themselves directly causo Inhorltud v.The..Mutagenicity'Issue (G-121 at 11); Genetic abnormalities are' . further significant in that they often . abnormalities, these flndlngs are A. Introduction affect an individual from the moment of nevertheless biologically significant for 1. Issue Presented: The second issue birth onward. rather than merely during two reasons: (a) because breaks may ,presented is: Whether the evidentiary later life, as-do-many other diseases. lead to another typo'of chromosome record establishes to a reasonable Thus, great caution should' be exercised aberration. called "exchange Ilgures," certainty that cyclamate does not cause in determining the mutagenic potentia! which do cause heritable genetic genetic damage and is not mutagenic. of cyclamate. diseases: and (b) because breaks may (ID at 4). In layman's terms, the question 2. Conclusion: For the reasons stated be indicators of gene mutations whloh is whether cyclamate causes changes in below; I find that Abbott has not shown may also cause inherited abnormnlltles the genetic code which could lead to an .that there-is a reasonable certaintythat (see Subsection F.2.c.belowl.I would! abnormal individual ih future , cyclamate does not cause heritable therefore categorize these six studies as generations (G-124-at7: see also A-800 genetic damage. being "inconclusive but suggestive" of at 1-;3). One expert defined mutagenicity 3.Summary ofEvidence:As discussed mutagenicity. These flndlngs, by most succinctly as the induction of in detail in Section II. above. the themselvea.require the denial of Abbott's food additive petition. "heritable genetic.damage" (G-12~at statutory scheme governing the 11). evaluation of a food additive petition CertifiedColorManufacturer'sAssn v, The basic genetic material in man, and provides that the petition shall be Mathews. 543F.2d 284,297 (D.C.Cir. animals is called deoxyribonucleic' acid deniedifa fair evaluation of the data 1976); accord. EnvironmentalDefense (DNA).The DNA, which forms the fails to establish that the food additive Fund v, E.P.A., 598F.2d 62, 89.(D.C.Cir. genetic code, is distributed among the will be safe under the specified 1978): Ethyl Corp. v, B.P.A., 541 F.2d1, many "genes" thatdetermine traits to be conditions ofuse. 2~ U.S.C;348(c)(3)(A) 37-38 (D.C. Cir:) (en bane), cert: denied, inherited. These genes are grouped in and 21 cm 170.3(i).In order for this 426U.S. 941 (1976): see Hercules,Inc. v. 1978)~ packages called "chromosomes." determination to be made. the parties .E.P.A., 598F.2d 91,110 (D.C.Cir. have introduced into the record 72 Moreover, a number of cytogenetic Chromosomes are physically much ~ larger than genes and can be seen under scientific studies'designed to test the studies performed in vitro provide a microscope. Humans have 46 . potentialmutagenicity of cyclamate and additional support for this conclusion chromosomes (23pairs) eachof which its metabolites; Ofthese, 49 studies (see SubsectionF.7. below). containsnumerous genes (A-800 atl-2).. were performed on live animals or The valid "negative" in vivo Either chromosomes or genes can,be human. beings (called in vivo). and the ' cytogenetic studies are insubstantial.by harmed in such a way as to create an remaining 23 studies were performed in comparison. Although four such valid abnormal individual in future test tubes using plant, anlmal or human , studies (A-143, A-151, A-716 and A-tlt1, generations (G-123 at 3: G-124 at 14: A cells (calledin vitro).The parties agree App. 19) found no statistically 800 at 1). The question before.me in this that the dispositive information must significant increase in chromosome proceeding is whether Abbott has come from in vivo studies because only . aberrations"each, of.these is readily produced evidence which proves that in these can the.testcompoundbe. distinguishable from the suggestive there is a reasonable certainty that examined'under conditions most closely findings just described. For example, cyclamate does not cause the-type of , approximating;actual humanuse one negative study (A-15i) used an genetic damage, either to chromosomes. [Abbott's Briefat44; Bureau's Briefat entirely different animal species (mice). or to genes, which may lead to an 73-74; G-124 at 30). The remaining three studies, although abnormalindividualin future- ' I find that the.results from one.group" using the same species. as one suggestive generations. - oiin vivo studies,Galled cytogenetic study (Chinese hamster), analyzed a From a medical standpoint; • studies. raise a-serious question as to. different type of celh the negative' mutagenicity is an extremely significant the potential.mutagenicity ofcyclamate. 'studies using bone marrow. (A-143) or • issue. Dr. Legator, the Bureau's chief . An in vivo "cytogenetic" study,. as will spermatogonial cells (A-716 and A-811, mutagenicity expert. explainedas be explained in greater detailin . App.19) versus blood.cells (G-45) for follows:. Subsection E.3.below, is.designed to. the suggestive study. (See discussion of measure a testcompound's effectupon individual studies in Subsection F.5. There are a variety of genetic diseases Down's syndrome, which is a product of chromosomes•. Sixin viva cytogenetiu below], Thus, none of the negative chromosomal abnormalities, various studies each found a statistically studies directly. rebuts, any of the neurological dlseases..mental retardation. a significant increase. in chromosome. suggestive evidence. host of inborn errors of metabolism; Genetic aberrations. These findings were I have eliminated from consideration abnormalities in our population are probably' obtained in five.different species: asqelng "deficient" 15 additional in _ the most signiflcarit health burden we now Holtzman.rats (G-9. both portions), sdvo cytogenetic studies because they 1\ - face. Indeed. it has been estimated that25' Mongolian gerbils (G-26). fetaIlambs do'not meet the minimum criteria set percenf of our overall health problems are ' (G-44). Chinese hamsters (G-45) and forth in Subsections C.2 and 3. below In due to genetic or genetically related'diseases. human. beings 0-1): and in three. terms of statistical or biological (G-124 at 7: see also G-122 at 7 and G- . different types of cells: bone marrow significance: (See discussion' of 121 at 11). Other Government agencies, (G-9 and G-26), blood (G-44. G-45, and individual studies in Subsection F.6 such as the EnvironmentalProtection J-l). and spermatogonia (G-9). (See' below). Agency. have also recognizedthe ' . discussion of individual studies in Mutagenicity studies on cyclamate medical significance of mutagenicity. As SubsectionE.4. bel~w.) ~. wex:.e also performed in three other typos Federal Register I Vol. 45. No. 181 I Tuesday, September"16.~1980 I Notices 61507 of in vivo studies: host-mediated assay; F.2d at 37-38; see Hercules, Inc. ". weights depending upon the nature and dominant lethal assay; and drosophila. E.F.A., supra, 598F.2d at110. Moreover. extent of any internal flaws. Although the findings in each of these the effect of "inconclusive but d. Deficient:A "deficient" study is groups were predominantly negative. suggestive" evidence on an agency's one which doe's not meet the minimum they do not outweigh the suggestive safety analysis applies with equal force criteria for either statistical orbiological cytogenetic findings just described to human health risks other than cancer. significance set forth below. Deficient because known mutagens have been Ethyl Corp. v. E.P.A., supra (lead studiesare entitled to no weight at all. found to show mutagenic effects in some poisoning). Thus, Abbott's food additive 2.Statlstlcal Significance. In contrast test methods but not in others (G-124 at petition must be denied ifa fair to the sharp debate on statistical 9-10 and 31;Tr. at 933-34;Tr. at 498-501; evaluation of the evidence suggests that significance sparked by the Tr. at 717-18 and 734).(See discussion of cyclamate may cause heritable genetic carcinogenicity data, the parties are in studies in Subsection G. below]. damage. For the reasons stated below, general agreement as to the statistical Finally. the record contains some that is precisely the situation here. significance of the mutagenicity studies. additional in vitrofindings. These types This is because the studies of studies. however. are never sufficient C.Criteria for Ole Evaluation of , predominantly reported findings at the to outweigh suggestive in vivo findings IndividualMutagenicityStudies. .05 confidence level; this was uniformly because in vitro studies. being I have adopted several minimum true, in fact, among the pivotal group of performed in test tubes, cannot take into criteria. involving both statistical and suggestive in vivo cytogenetic studies. .account a live animal's metabolism (Tr. biological significance. necessary for a Thus. the only issue relating to at 937-38; see also G-124 at 10; G-121 at study to be considered valid. Based statistical significance of the 12). (See discussion of studies in largely upon these criteria. mutagenicity mutagenicity studies is whether a Subsection G.5.below). studies may be clasified into four statistical analysis had been performed In sum, the repeated in vivo categories: (a) positive; (b) suggestive of on a given study. I believe that the cytogenetic findings of breaks raise a a mutagenic effect; (c) negative; and (d) performance of a statistical analysis is a serious question about cyclamate's deficient I first will define these terms minimum requirement necessary to mutagenic potential-specifically. its and then set forth the minimum criteria. demonstrate the validity of a study's capacity to induce exchange figures and 1. Classificatlon ofMutagenicity results. Those studies which fail to give gene mutations. both of which are Studies. I have adopted the same this critical information (and where the capable of producing genetic classification terminology for the parties have not themselves performed a abnormalities in future generations. mutagenicity studies as I used for the .statistical analysis using reported data) Indeed. as Dr. Legator concluded. "Any llarcinogenicity studies. Very briefly, have been eliminated from compound which shows the [mutagenic] these terms are defined as follows: consideration as being "deficient" (see effects cyclamate has shown should be a. Positive: A "positive" study is one »e- A-217J. considered a high risk agent" (G-124 at which conclusively demonstrates that 3. BiologicalSignificance: I have also employed three minimum criteria 26). Given this strongly suggestive cyclamate causes heritable genetic mutagenicity evidence. the statutory necessary to establish the biological damage. There are no such studies on significance ofa study. These involve: a) scheme mandates tliat Abbott's food this record. additive petition be denied on this study size; b) reporting of data; and c) additional ground. b. Suggestiveofa MutagenicEffect:A positive controls. "suggestive" study is one which. a. Study size. For a study to be B. TheStatutoryScheme although inconclusive. suggests that considered valid. itmust employ a I have already discussed the cyclamate may cause heritable genetic sufficient number of animals to give the legislative history as well as the judicial damage. The principal examples on this study an adequate degree of sensiti\ity. and administrative interpretations of the record are the in vivo cytogenetic - Dr. Legator testified, for example, that general safety clause (see Section II. studies which found a statistically \'oithrespect to the in vivo cytogenetic above), and I adopt that discussion here. significant increase in chromosome studies, at least ten animals should be . Nevertheless. one point worth repeating aberrations, predominantly breaks. used per treatment group (G-124 at18). here is that the general safety clause These studies are suggestive rather than This figure was based not only on Dr. requires disapproval of a food additive positive because breaks themselves are Legator's own experience, but also upon petition ifthe evidence "suggests" lack not inherited. Rather. as explained in the minimum protocol recommended by of safety. even if that evidence is Subsection F.Z.c.below. breaks are the Ad Hoc Committee ofthe inconclusive. For example, in Certified biologically significant because they Environmental Mutagenic Society (id.). ColorManufacturer's Assiz v. Mathews, may: (a) lead to exchange figures; andl Abbott produced no expert testimony to supra, which presented an analogous or (b) be indicators of gene mutations, the contrary. I have therefore adopted situation involving the act's Color both of which are capable of inducing the figure often animals per treatment . Additive Amendments of 1960. the court heritable genetic abnormalities. (Note. group as a general guideline in stated: however, that findings of breaks atP<.05 determining the adequacy ofanin vivo are termed "positive findings" even citogenetic study's sensitivity. (In so The information available to [the though those studies are termed Commissioner of Food and Drugs] indicated a stating, however. I note that all studies statistit resolved them on a study-by-study bais was not translated, and is therefore, . positive control to be the determinative by considering the expert testimony on impossible to evaluate. factor in declaring the study to be that particular study. . , . c. Positive Controls. Ideally, every "deficient" (see, e.g., A-274 in This criterion of minimum study size mutagenicity experiment (except those Subsection F.6.a. below). has a different effect on "negative". conducted on human beings) should The lack of a positive control has 11 studies than it does on "suggestive" have a positive control group, which is different effect upon a study with studies. The concept of adequate simply an additional treatment group "positive flndings" (e.g., suggestive in sensitivity means that the study must dosed with a known mutagen, (see Tr. at vivo cytogenetic studies). Statistically have been large enough [i.e., sensitive . 717). As the parties agree, the purpose of significant (P<.05) flndlnga in tho test enough) so.that, ifa test compound is a positive control, is to serve as a check group are sufficient, by themselves, to mutagenic, there is sufficient likelihood on "thesensitivity of the test-i.e., to _. demonstrate that the sensitivity of the that the mutagenic effect will be ensure that the experiment is able to experiment is adequate (Tr. at 975). As detected. Thus, ifan in vivo cytogenetic detect a mutagenic effect where one Dr. Legatee tes'tified "Ifone gels a[n] study using only three or four animals .would be expected as to bapresent [Tr effect without a positive control, thai per group produced "negative" findings, at 717;975). Inlaboratories which again, as I said can be classified as a no confidence can be placed in those . specialize in the particular type of good experiment" (id.). Thus, I have. mutagenicity testing being performed, considered these studies to be valid results. For this reason, I have rejected e.g.~ as being "deficient" these so-called this same assurance can be gained from (see, G-45 in Subsection F.4.e "negative" studies. positive control data derived from below). previous experiments. Such data are In contrast, ita study with only few D. Criteria fbr the Evaluation of a .called "historical controls" (Tr. at 960). animals produces statistically Mutagenicity Evidence as a Whole significant results, it cannot be criticized The following examples illusrate how results from positive controls either ~ter each study has been reviewed' for being too insensitive. Quite the verifY, weaken, or completely nullify a and classified, the evidence as a whole contrary, what such a result suggests is study's otherwise "negative" findings. must be evaluated to determine if that the test compound is sufficiently . First, ifthe positive control values are Abbott has demonstrated that there is 11 potent that it is capable of being clearly positive, this verifies the reasonable certainty that cyclamate detected by even an insensitive study negative results from the test compound does not cause heritable genetic (Tr. at 941). I therefore consider results and enhances their credibility because damage. This overall evaluation from studies in this category. the experimenthas been proven to be necessarily involves a judgmental (particularly G-44) to be facially valid, able to detectmutagenicity where itis process by the declslon-maker, although perhaps entitled to slightly less expected to exist. In contrast, ifthe especially in making factual weight than results derived from a large!" positive control values are positive but determinations such as whether certain test population. This approachis below their norm, the test compound's negative studies outweight other consistent with that taken in the "negative" findings are of questionable suggestive ones. To objectify this carcinogenicity section of this decision. significance because the experiment has process as much as possible, however" (See discussion of the calcium been shown to be not as sensitive as it the record contains three criteria. Thene cyclamate portion of the first Friedman " should be (see discussion of A-7.16 and involve the necessity for using a battery study, Section IV.B.2.b.(1)(c) above.) A-811, App. 19 in Subsection F.5. of test methods, for testing different b. Reporting ofData. The second below). Finally, ifthe positive control animal species, and for obtaining results criterion under the rubric ofbiological values are negative, this completely from different laboratories. significance is that each valid study nullifies the test compound's negative 1. Battery ofTest Methods: The must contain an adequate presentation -results because the experiment has been parties are in agreement that cyclamate of data so as to enable a full evaluation shown to be too insensitive to detect a must be tested in wide variety of of the study's results. For example, some known mutagen (see discussion of bone experimental methods because know of the studies are published only in marrow portion of A-177 in Subsection nutagensoften produce posltlve results "abstract" form, a mode which normally F;6.a. below), in some test methods but negative contains only a brief summary of the Where a negative study contains no results in others (G-124 at 9-10 and 31; study's methods and results and positive control data at all, I have taken Tr. at 933-34; Tr. at 498-501; 'fro at 717... virtually-no presentation of data. the following approach. First, for . 18 and 734). As Dr. Legator explained: Addtionally, other scientific papers negative studies with no internal flaws At the present state of the art, all ofour contain results of several types of suggesting the experiment's lack of tests for describing or characterizing studies that were run concurrently; in sensitivity (i.e., A-143 and A-151 in mutagenicagents have very serious these, the results orone portion (e.g., the. Subsection F.5. below), I have not drawbacks. Often,a partlcular typoof test , in vivo cytogenetic portion) were consideredthe absence ofa positive may miss a particular agontbecause of tho apparently of secondary interest to the insensitivityofthe proceduro, or the type of control, by itself, to render the study chemicalbeingtested, the timeof analysis.or investigators and therefore insufficiently "deficient." This is because the many otherfactors.The groatmajorityof presented. I have rejected as being investigator may have had historical compounds. with very fowexceptions, do not "deficient" all of these studies, positive control data which was not giveus a 'positive effectin all tests. Thoroforo. regardless ofwhether they reported reported in the published paper (fol' all responsibleagenciesin this area positive or negative findings, which do example, compare G-9 with Tr, at 960). recommend that we use a battery of teatil, not supply enough information to be In this situation, I have treated the lack that is. a numberof tests, to study a slnule assessed intelligently (see e.g., G-124 at of a positive control as reducing the agent. 19; see also. Tr. at 490 and G-122 at 21). I . weight to be given to a study, rather (G-124 at 9). Dr. Legator goes on to clle have also not considered the results of . than as affecting its'overall validity. In two examples of known.mutagens one purportedly negative in vivo contrast, where other factors-ln a study (ionizing radiation and nitrogen . cytogenetic study (A-241) because it suggest that the test is insensitive, r mustard) which do not produce poattlva was submitted in ~ foreign language, have considered the absence of a . effects in one or more accepted .' Federal Register I Vol. 45, No. 181 I Tuesday, September 16. 1980 I Notices 61509 mutagenicity test methods (id. at 9-10). 124:Tr. at 894-976).I have reviewed cyclamates and their metabolites cause Dr. Green. another Bureau witness. each expert's curriculum vitae and a significant amount of chromosome cities five additional examples of this relevant testimony and find that each damage" (A-BOO at 7) (emphasis added). type [Tr, at 498-501).Abbott's chief holds outstanding qualifications in the This statement has two majorflaws. mutagenicity witness, Dr. Hsu, also field of mutagenicity testing. Both men First, Dr. Hau's conclusion is vague. He agrees with this general proposition (Tr. have extensive experience in the design talks in terms ofa "significant amount" at 717-18 and 734).Thus, for Abbott to and execution of both in vivo and in of chromosome damage without be able to establish safety, it must vitro mutagenicity testing, using a elaborating on how much or what kinds present a complete battery of variety of test compounds. Moreover, of chromosome damage would be mutagenicity tests which show negative both men have special expertise in the needed Wore he would term it results. area of cytogenetics which is of central "significant." Second, Dr. Hsn's 2.DifferentAnimal Species: importance in this proceeding. conclusion is incomplete. He mentions Mutgenicity tests must also be Dr. Hsu gained his experience in only the potential for chromosome performed in a variety of animal species academia, primarily at the Univcrsity of damage without offering any opinion on in order to prove safety. This is because Texas (Houston campus), but also at the potential for gene mutations. This a compound may produce negative Baylor University, Brown University, omission is somewhat surprising inlight results in one species but positive Rice University, and Wayne State of his own introductory statement that a results in another (Tr.at 965:A-143 at University. Dr. Legator served ten ycars "(gene) mutation affecting an important 16: see also G-123 at 5). Although it (1962-72) as Chief of the Genetic gene cancause lethality; and a mutation would be impractical to require tests in Toxicology Branch of the Food and Drug , affecting a less important gene can alter every imaginable animal species, Administration. Dr. Legator then moved the organism's morphology or studies designed to rebut specific on to academia, first at Brown physiology" (A-800 at1).:15 positive or suggestive findings should be University and most recently at the I find these shortcomings of Dr. Hsu's performed using the same animal Univcrsity of Texas (Galveston campus), testimony to be extremely significant species and strain {id} Thus. the species which is the same university where Dr. and of far greater consequence than are employed is an important factor to be Hsu teaches. Indeed, Dr. Hsu was one of Abbott's criticisms of Dr. Legator. considered in weighing the suggestive the cytogenetic instructors at a Abbott suggests. for example, that Dr. findings against the negative ones (see toxicology course organized there by Dr. Legator has ''preconceived notions"· discussion in Subsection F.1. below). Legator (Tr. at 918).The two men about and an "emotional involvement" 3.DifferentLabaoratories: The need therefore know each other well, and in the cyclamate issue and therefore is to obtain test results from different each has readily acknowledged the not able to render an objective. laboratories has recently been other's expertise (Tr. at 719;919). scientific opinion on this subject scrutinized and documented. AccoPding (Abbott's Brief at40;Tr. at907).Abbott to Dr. Legator, eight laboratories Rather than training and experience, the pivotal factor in the relative bases this claim on certain letters to the performed a collaborative in vivo press (G-136) and to FDA (A-828 and credibility of each man's testimony is cytogenetic study on the compound A-83a) in which Dr. Legator advocated a the extent to which each has triethylenephosphoramide (TEPA), a cyclamate ban (Abbotrs Briefat40-41). demonstrated a familiarity with the known mutagen. Before commencing the I have reviewed these letters and actual study, the investigators agreed on cyclamate studies of record In this decline to find the inferences which standardized procedures and on proceeding. Dr. Hau's entire direct Abbott suggests. The first letter was standardized definitions for scoring testimony evaluating the cyclamate based, from a mutagenicity standpoint. slides. Nevertheless, one of the eight evidence consists merely of a brief, on objective scientific dll.ta-i.e., the laboratories produced results that were general summary which does not even then recent laboratory findings of Dr. clearly different from the other labs (Tr. mention the name or author ofa single Legator (G-9) (see G-136 atRef,8). at 923-25).Thus. the extent to which the cyclamate study (A-BOO at 7-8). Instead Moreover, Dr. Legator's views in this key data comes from the same or of criticizing these cyclamate studies, letter were shared by four other different laboratories is also a factor to Dr. Hsu seems to rely on a review article prominent scientists, including a Nobel be considered in weighing the evidence. which is not of record in this proceeding Laureate [Tr. at907).The thrustof the and whose completeness is in some E. Credibil{ty ofExpert Witnesses other two letters is that insufficient doubt [Bureau's Brief at 108;Tr. at 919 information was then known about the The credibility of the expert 21).Moreover, the fact that Dr. Hsu has way cyclamate is metabolized to permit mutagenicity witnesses is very much in not conducted any cyclamate studies a responsible finding that cyclamate is issue (see Abbott's Brief at 39-42 and himself (Tr. at 732)precludes another safe. This opinion also was grounded in Bureau's Brief at 106-110).In reaching avenue by which he may have become scientific facts which were stated in the my own conclusion as to the credibility familiar with all or part of the cyclamate letters themselves (A-828 and A-830). of each expert, I have considered the evidence. In contrast, Dr. Legator Abbott also suggests that these letters following factors: (1)his training and described and evaluated in some detail reveal Dr. Legator's extra-scientific experience: (2) the extent to which he many of the cyclamate studies, opinion that "cyclamate should not be has demonstrated a familiarity with the especially the key cytogenetic ones (G cyclamate studies of record: (3) the . 124 at 16-25). Thus, Dr. Legator's -In any tvtnt. Dr.HR'. COOcIlISiou wouldDOl extent to which his testimony is testimony on specific studies is entirely IUpporta fiadIn& orsafety.~ discuued above.the corroborated or supported by other unrebutted by Dr. Hsu (or any other gtntral Nfety cIauaeof the act requiresdisapproval ... evidence in the record: (4) clarity or Abbott witness). of a foodaddiliYe ~ a sbowingof eYidence "IUfaeIUve" oC_tqenicity.even II that mdence vagueness of his opinions: and (5) In addition to Dr. Hsu's questionable II Inconcllllive. CArorltldColorMOlIufoclurers possible bias. familiarity with the record, his Assn v. MaIMWToIUPl'C1. 543F.2dalm;occord: The parties' chief mutagenicity conclusions on the ultimate Enl7rollJ1HUllaJDe[etlltt Fundv. U.A.. supra.598 experts are Tao-Chiuh Hsu, Ph.D. for mutagenicity issue are not convincing. F.2datts;EthylCotp. v.UA. supra.5-11 F.2dat 37-38;.~ Hercum. Inc. Y. U.A.. svprrt. 5fl8F.2dal Abbott (A-BOO; Tr. at 715-734)and Dr. Hsu concluded that "there is no lID. Thus. thefilctthat "decisive"evidencedoes not Marvin Legator, Ph.D. for the Bureau (G decisive evidence to show that e.'dsldoesIIOtmean that AbbottshouldpreloCliL 61510 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices = approved as a food additive because it and Zimmering are all credible hamsters. Finally. Bauchlnger, et al, II-l) provides no benefits to society" witnesses.' found a statistically significant Increase (Abbott's Brief at 41).This is not the I also need not discuss in detail the in breaks after analyzing the blood cello import of these letters. Dr. Legator was qualifictions of Abbott's other • of human test subjects. simply stating that cyclamate possesses mutagenicity expert, pro Lorke (A-811T Abbott has challenged the valldity of no societal benefits capable' of . 'and A-827), but for a different reason. each of these six studies. In the study outweighing the public health risks Unlike the other witnesses, Dr. Lorke by-study analysis which follows later in which he perceives. Under the statutory did not present either written or live this decision. I have considered each of scheme, however, possible societal testimony in which he evaluated specific these alleged flaws and have concluded benefits are not even to be considered. studies or the evidence as a whole. that each of the studies has strengths . See 21 U.S.C.348(c)(3)(A). Since Dr. Instead, Dr. Lorke merely attached to his (e.g., the dose response in G-9, both Legator clearly distinguishes between curriculum vitae a number of cyclamate portions. G-26 and G-44) which the "risk assessment" and the "benefit mutagenicity studies which he . outweigh any claimed weaknesses. assessment" in hisanalysis, I have performed. These studies were properly Moreover, the fact that findings of considered the former but not the latter introduced into the record and have breaks were reported by five different in my evaluation. been reviewed along with the other laboratories using five different animal experiments: and-the general statements species and three different types of cella Finally, Abbott challenges Dr. . made in them are entitled to the same greatly enhances the studios' collective Legator's scientific abilities by asserting weight as those of any other investigator that his positive findings with cyclamate h tud' f d credibility andmakes the evidence aq a In IS 0 (espeCllallv y m O«u-9, hi's most Important w ose. s y . . recor . . . . whole surpass the sum of its parts. Even study) were artifacts.since they could F. Evidence RmSIn.g.a SerIOUS QuestIon more important, Bauchinger's flndlnga not be replicated by other investigators' -. as to.the !-'futagenlCIty ~f Cyc~amate: from human beings lend confidence to (Abbott's Brief at 40).I disagree with The In VIVO CytogenetIC Studies. the extrapolation of the animal study Abbott for two reasons. First, Dr. 1. Summary ofEvidence.An in vivo findings to potential human use. Legator is not the only investigator who cytogenetic study is an established type The four valid negative studies oro has found an increased incidence of of mutagenicity test, carried out on ·live Brewen, et a1. (A-143). Cattanach, et al, chromosome aberrations in an in vivo animals, used to examine a test (A-151) and two studies by Larke. et al, cytogenetic test. As illustrated in compound's possible effects on (A-716 and A-811, App. 19).All of thooo Subsection F.4. below, four other chromosomes. The current studies reported no statistitally investigators, unaffiliated with Dr. administrative record contains 25in significant increase chromosome aberrations in treated anlmals over the Legator. have also reported a vivo cytogenetic studies. A review of st~dy statistically significant increase in each of these studies has revealed that negative controls. Brewen's used chromosome aberrations after using six are suggestive of mutagenicity, four the bone marrow of Chinese hamsters. cyclamate or CHA. Moreover, the are negative, and 15 are deficient. Cattanach analyzed spermatocytes of "negative" studies relied upon by The six suggestive studies (G-9 [both .mice, while both Larke studios involvod Abbott as demonstrating lack of bone marrow and germ cell portions], . the spermatogonia of Chinese hamsters. replicability of G-9 (i.e., A-177, A-195 G-26. G-44. G-45 and 1-1) collectively In comparing these negative studios and A-297) are all "deficient" in terms present strong evidence that cyclamate with the suggestive ones, it Is clear that of design or procedure and therefore do or CHA: causes chromo~ome the suggestive fmdings predominate. not detract from Dr. Legator's findings. aberra~ons. These findin~s were Cattanach's study is clearly (See dicussion in Subsection F.6. below.) predommantly breaks, with some distinguishable because he used an entirely differerit animal species I therefore find that Dr. Legator has evide~ce o! exchange ~gures. As fully (mouse). the three negative Chineso provided credible expert testimony explam:~ in the followmg Subs.ection of which was based on objective facts, not ~s .deClsIon, breaks are.blclogically hamster studies are also distinguishable personal bias. Moreover. the fact that he significant because they. (a) may lead to because they analyzed either bone marrow (A-143) or spermatogonia (A provided extensive detailed analyses of exc~ange figures: and (b} may be lfic di .' indicators of gene mutations. Both 716 and A-811, App. 19) cells whereas many of ~e sp:c .cstu es at Issue exchange figures and gene mutations are the positive Chinese hamster otudy (G , makes t!lls testimony far more capable of causing heritable genetic 45) examined blood cells. Thus, none of persuasive than the general, conclusory defects. the negative flndlngs directly rebuts any remarks offered.by Dr. ~su. The six suggestive studies may be suggestive study, . I need not go Into detail on the 'summarized as follows. Legator, et al, Moreover, each of the four negative e~pertise of the Bureau's oth?r (G-9) found a statistically significant studies has internal flaws which reduce Witnesses, Drs. Green. Epstem and increase in chromosome aberrations. the weight accorded to it. Brewen (A Zimmering, (G-123, G-121 and G-122, predominantly breaks, in both the bone 143) did not specify the size of his test respectively). as their credibility in not marrow IJIld spermatogonia o'I-Holtzman population, and neither he nor directly challenged by Abbott. I have rats. Majumdar and Solomon (G-26) Cattanach (A-151) supplied positive reviewed each's curriculum vitae and found similar results in the bone marrow control data. As to the Lorke studies (A relevant testimony and find that each , of Mongolian gerbils. Turner and 716 and A-811, App, 19), the positive qualifies as an expert in mutagenesis. Hutchinson (G-44) found a statistically control values for each were low. and Although each's testimony is limited in significant increase in both breaks and the test population size for the latter one scope (Drs. Green and Epstein to certain exchange figures (scored separately) in (A-811, App. 19) was too small (six per in vivo cytogenetic and dominant lethal the blood cells of fetal lambs, while van group) which lessened its sensitivity. studies, and Dr. Zimmering to Went-de Vries (G-45) found a I therefore find that the in vivo drosophila experiments), each of these statistically significant increase in cytogenetic evidence, when viewed as Q experts demonstrated a familiarity with chromosome aberrations (breaks and whole, strongly suggests that cyclamate .the specific studies evaluated. I exchange figures being grouped may be capable of inducing heritable therefore find that Drs. Green. Epstein together) in blood cells of Chinese genetic damage. This evidence alone in Federal Register I Vol. 45. No. 181 I Tuesday, September 16, 1900 I Notices 61511 sufficient to deny Abbott's food additive gaps. However, the comparison made in I also find that one in vivo cytogenetic petition. that study, when read in context. is study in the record (Turner and 2. Biological Significance ofDifferent slightly different from that presented b~' Hutchinson, G-44] reported a Types ofChromosome Damage. The Abbott. A-239 is a collaborative in vivo statlstieally significant increase in major issue surrounding the in vivo cytogenetic study conducted by four exchange figures after dosing fetal cytogenetic evidence involves the independent laboratories. (The study did Iambs with eRA. [See general biological significance of three types of not involve cyclamate or its discussion of this study in Subsection chromosome damage: breaks, gaps. and metabolites.) One purpose of the study F.4.d. below). Although this finding is exchange figures. was ". " .. to test the variability in not byitself sufficient to prove that a. Types ofChromosome Damage: As interpreting [jointly prepared] slide cyclamate is mutagenic. I believe this noted above•.deoxyribonucleic acid preparations by participants in their evidence does cast doubt upon the (DNA) is the basic genetic material in respective laboratories" (A-239 at 338]. safety of cyclamate in this regard. man and animals. The DNA, which This type of study recognizes the fact c.Bio!ogical Significance ofBreaks: forms the genetic code. is distributed that the cytogenetic analysis is, to some The central cytogenetics question. and among the many "genes" that determine extent, a subjective art as much as an one on whiCh the parties strongly traits to be inherited. These genes are objective science. The investigators disagree, concerns the biological grouped in packages called therefore sought to compare analyses by significance of breaks. This issue is of "chromosomes." Humans have a total of different laboratories of jointly prepared central Importance because breaks were 46 chromosomes; lower animal species, slides. These investigators. which the predominant finding throughout the •such as rats and mice. have fewer (A included Dr. Legator, concluded: six suggestive studies relied upon by the 800 at 1-2). Bureau [G-9 [both portions]. G-26, G-44. Physically, each chromosome (at the The results indicate that gaps are theleast conclusivecriterionin detennining G-45, and 1-1; see in Subsection F.4. cell cycle stage called metaphase) cytogenetic effects.The variabilitybetween below). contains two rods which are joined laboratories was greatest forgaps:in The Bureau's position is that breaks either at their centers or at one end, addition. the values forgaps resultingfrom are biologically significant for three depending on the species. Each rod is TMPand the differentdoses of DDTwere not reasons: (a) they lead to exchange • . called a "chromatid." Sometimes part of significantly differentin the ip [injected] and figures; (b] they are indicators ofother one of the chromatids cuts off and oral parts of the experiment.Agreement types of genetic damage such as gene separates from the main rod. Ifthe between laboratories was close for the mutations; and [c) they can cause separation is greater than the width of a criterionofbreaks, and was even closer for heritable genetic damage themselves chromosome, the aberration is called a rearrangementfigures • " • (Bureau's Brief at100(03). "break." Ifthe separation is less than (A-Z39 at 349-50]. This passage, read as Abbott does not directly respond to the width of a chromosome, it is called a a whole, clearly demonstrates that the points [a) or (b). but does strongly "gap" (G-124 at15; Tr. at 958).These are investigators rank-ordered the three" disagree with the third. arguing that the definitions published by the Ad Hoc types of chromosome damage in terms chromosomes with just breaks are not Committee ofthe Environmental of the agreement/variability between inherited because they either repair Mutagenicity Society (id). and the ones themselves or die (Abbott's Exceptions which I adopt laboratories. In other words. when four at 36; 47; Abbott's Brief at 47-52; 56-57; Dften chromosomes will repair laboratories separately eeadjointly themselves after 'breaks or gaps have prepared slides, their Iindlngswere and 61). occurred, or else die. Sometimes, "close" for breaks, "even closer" for The ALJ did not make specific however, two different chromosomes, exchange figures. and varied the findings on this issue but did note that each with breaks, join together at the "greatest" for gaps. I interpret these statistically significant findings of site of those former breaks. When this results to mean that in an in vivo breaks could not be disregarded (ID at happens. the resulting configuration is cytogenetic study, findings of breaks can 34). called an "exchange figure" (A-800 at 4; .be considered reliable, exchange figures A careful review ofthe testimony on G-123 at 3; G-124 at 14). Exchange even more reliable, but gaps not very this issue shows that it strongly supports figures are also sometimes referred to as reliable. In so finding. I note that this the Bureau's position that breaks are "reunion figures" (A-l77 at Table III), conclusion only reaches the issue of biologically significant for the first two "rearrangement figures" (A-297 at Table whether certain findings in in vivo of the three reasons advanced by the 1), "translocations" (A-716 at 243). or cytogenetic studies can be considered to Bureau. I will now discuss them in the "major structural aberrations" (G-44at be accurate. not whether those findings, order presented by the Bureau. Table 1). (For pictorial illustrations of even if accurate, are biologically (1] As Leading to Exchange Figures: breaks. gaps. and exchange figures. see. significant. For the remainder of this As Dr. Green explained: e.g; G-45 at Figure 2 and A-239 at Subsection. I will consider the latter E.'Cchange figuresare producedas a result Figure 2). issue. of the rejoiningof chromosomes which Abbott questions the comparative b. Biological Significance ofExchange poness chromatidbreaks. Itis. therefore. biological significance of breaks versus Figures: The parties agree that exchange apparent that chromatidbreaks are the gaps versus exchange figures (Abbott's figures are biologically significant in necessary e...ents which subsequentlylead to Brief at 61).Relying upon A-Z39 at 350, that they can survive and pass on exchangefigures. Abbott claims that"'" " " gaps are the genetic defects to the next generation. (G-123 at 3; see also G-124 at14; A-800 least conclusive criterion in As one Bureau witness. Dr. Green. at 4]. This theoretical point is confirmed determinating [sic] cytogenetic stated: "It is generally thought that by the slatistically significant findings of effects • " "'. and chromatid breaks are exchange figures are the type of exchange figures in the Turner and only slightly better • " • the best abnormality that can be associated with Hulchinson study (G-44]. Moreover, criterion • • • are rearrangement heritable genetic damage" (G-123 at 3: according to the experts, the fact that figures " * .11 (Abbott's Brief at 61). see also G-124 at 14; Bureau's Brief at several studies found breaks without I agree with Abbott that A-Z39 rank 100-02; and Abbott's Brief at 49]. I agree exchange figures is not unusual. As Dr. orders excbange fixtures. breaks and and find accordingly. Legator explained: "••• the frequency 61512 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices , . here of exchange figures by our current already been reported. I recognize that correlation between breaks and gene techniques are always minimal. That is, thisis a heavy burden to impose upon mutations. On this record, however, we do see it, but it is the rare kind of -Abbott, but itis one that I believe is there is oilly one gene mutation study event" (Tr. at 956:see also Tr. at 957and necessary to prove safety, as the statute conducted on mammalian cells, and that G-124 at 29-30).Dr. Legator attributed requires. • was an incompletely reported in vitro the low frequency of exchange figure (2) As Indicators ofGeneMutations: study conducted by Chu, et al, (A-600: findings to the imperfect state of the art Breaks are biologically significant for a G-47). Chu reported negative findings for this type ofscientific test: "•.• when second reason, the Bureau argues, with cyclohexylamine (CHA, a we do see breaks or gaps, ifwe readjust because they serve as indicators of metabolite of cyclamate) and our techniques or timing, we probably other types of genetic damage, positive fmdings with N· could see exchange figures as well" (G especially gene (or point) mutations. Dr. hydroxycyclohexylam1ne (N-OHCHA, 124 at 14: see also G-124 at 30 and G Hsu, Abbott's chief witness, described another metabolite) both with Chinese 123 at 3). As Dr. Legator concluded: "It is genes and gene mutations as follows: hamster cells. These findings, however, very difficult to find a compound that " A cell of an organism contains numerous were reported only in a biref abstract. has been thoroughly investigated that genes each of which determine a particular with no supporting data, and are does not cause exchange figures when step of a-biochemical process. Itis a therefore entitled to little, ifany, weight breaks or gaps are found" (G-124 at 15: sequence of DNA with code which (see general discussion on abstracts in see also Tr. at 957). determines a particular protein. H the code Subsection C.3.b. above and specific On the basis of this evidence, which is changes in whatever manner or ifthe code is discussion of the Chu study in unrebutted by Abbott's chief expert, Dr. missing, the gene becomes 'mutated' or Subsection G.5. below.l I therefore Hsu, I find that one basis for the deleted respectively, and the gene cannot conclude that Abbott has not shown biological significance of breaks is that perform its designated function. A mutation affecting an important gene can cause that there is reasonable certainty that they will likely lead to the formation of lethality; and a mutation affecting a less cyclamate or its metabolites do not exchange figures which, as the parties .. important gene can alter the organism's cause gene mutations. agree, cause heritable genetic damage, morphology or physiology. (3) As Heritable GeneticDamage In theory, Abbott could rebut this Themselves: Finally, the Bureau conclusion by presenting sufficient valid (A-800 at 1). Thus, in terms of heritable damage, gene mutations are as -' contends that breaks "can cause serious negative studies proving that there is a genetic damage themselves" (Bureau's reasonable certainty that cyclamate or biologically significant as exchange figures (G-124 at 8). Brief at 102).Specifically, the Bureau its metabolites do not cause exchange argues that chromosomes with breaks figures. Abbott believes that it has Dr. Legator testified that there is an "extremely good" correlation in other can be replicated and passed on to already done this (Abbott's Exceptions progeny, and, relying on Dr. Hsu's at 65, 72, 73 and 74-75: Abbott's Brief at compounds between chromosome, abnormalties and gene mutations (Tr. at testimony, this constitutes genetic 56-63), but I disagree. Although Abbott damage because it represents a change introduced into the record 18 studies 931)."[I]n fact," he said, "I cannot think which it labeled as being "negative," I of more than perhaps one exception-[of in the genetic code (id. at 102-03). have found 14 of these studies to be compounds] that cause chromosomal Abbott responds that breaks "deficient" because they do not meet the damage that do not also cause gene themselves are not biologically minimum criteria for a valid study (see mutations" (id; see also G-124"at8: G significant because they will either discussion in Subsection F.6. below). 123 at 3). Dr.legator has also repair themselves or die; thus, a single Of the four valid negative studies, emphasized this correlation between broken chromosome, when not part of each was internally flawed because it chromosome damage and gene an exchange figure, will not be inherited lacked validation by positive controls mutations in a context completely by future generations (Abbott's (A-143 and A-1S1), had unusually low independent of cyclamate: "Although Exceptions at 36;44;Abbott's Brief at positive control values (A-716 and A there is no proven quantitative 47-52; !i6-57). 811,App. ~9), did not specify the size of relationship between point mutation and I agree with Abbott that chromosomes (A-143) or did not employ a sufficiently chromosomal changes, the correlation with breaks themselves do not large (A-811, App. 19) test population. between either physical agents or constitute heritable genetio damage. A The best evidence presented by Abbott chemical agents that can cause both review of the passages from the attempting to demonstrate the lack of types of alteration has been well . testimony which the Bureau quotes in its exchange figures is the Cattanach study established [reference omitted]" (A-239 brief shows that, when read in context, (A-1S1).As explained in Subsection at 349).At least one other investigator of they so not support the Bureau's F.5.b. below, that study was conducted - record in this proceeding agrees: ". • • position. For example, Dr. Legator exclusively to look for exchange figures. minor chromosomal lesions which atressedr'The intriguing thing about However. Dr. Cattanach conducted his cannot be regarded as permanent " exchange figures as opposedto breaksis study on mice which was not one of the breaks, may well be indicators of that these rearrangements can survive species in which evidence of breaks submicroscopic damage, pinpoint and multiply" (G-124 at 14) (emphasis were found (i.e., rats, gerbils, lambs, mutations" (Schoeller, G-::i8 at 3j. added). This statement clearly suggests Chinese hamsters and humans), Therefore, on the basis of this that chromosomes with just breaks For Abbott to prove to a reasonable unrebutted expert testimony, I find that . themselves cannot "survive and certainty the lack of exchange figures, it breaks are also biologically significant multiply," I read the language quoted by would have to present a group of valid because they serve as indicators of gene the Bureau to mean that chromosomes negative sfudles designed to detect mutations which, as the-parties agree, with breaks which do not die (or exchange figures; these studies should can cause serious heritable damage. correctly repair themselves) are have large test populations, several dose In theory, Abbott could rebut this significant because they may lead to levels, and validation by positive . expert testimony with valid negative exchange figures. The Bureau's controls. Moreover, these studies should gene mutation studies demonstrating reference to Dr. Hsu's description of include the animal species in which that cyclamate is the exception, rather gene mutations is also, I believe, not positive findings of breaks in vivo have than the rule, with respecUo the _ supportive. As noted above, Dr. Legator Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61513 was quite clear that breaks are merely specimens;and (3) analyzing cell (These cells are called "metaphases" "indicators" (as opposed to direct specimens. because the cells are at the metaphase . causes) of possible gene mutations.I In the dosingstage, the test group of stage of the cell cycle.)Although the therefore conclude that, on the basis of animals or humans is given the test pubUshedreport of the study does not so the current record, breaks themselves do compound, either by feeding or state, these sUdeswere coded so that not constitute heritable genetic damage. injection,for a specified period of time. the persons reading them had no d. BiologicalSignificance ofGaps: Concurrently, both negative and positive knowledge of whether they came from The biologicalsignificanceofgaps is a control groups are usually identified and treated or controlgroups (Tr.at 960). less important issue for the purposes of dosed by the same means and for the (2)StudyResults: Analyses of the this proceedingbecause the suggestive same duration.The negative control bone marrow cells revealed a in vivo cytogeneticfindingsrelied upon group is given a placebo, and the statistically significantincrease [p<;.01) by the Bureauwere based primarily _ positive control groupis given a known over the control group in the percentage upon breaks rather than gaps (G-9 at mutagen. of cells,with breaks in each of the four 1140; "singlechromatid breaks At the conclusionof the dosingperiod, highest dose groups.Moreover,a linear predominated"; G-26at 191, 193: breaks cell specimens from each group are dose-response was observed throughout scored separately; G-44 at 409: breaks .obtained for microscopicanalysis. Three these four groups.The authors also • scored separately; G-45 at 417: types of cells were used in the noted "infrequent exchange figures" (G exchangefigures, breaks and gaps all cyclamate experiments:bone marrow, g at 1140), but presumably, these alone groupedtogether;and J-1 at Table 3: blood, and sperm cells, Obtainingbone did not reach statistical significance. breaks scored separately.) marrow cells sometimesrequires that The ALJfound that this study Nevertheless, the Bureauargues that the test animal be sacrificed.This produced ''positive results" with a gaps are entitled to some weight procedure, therefore,is usually reserved "dose-response" trend (id. at 25-26. 35). (Bureau'sBriefat 10:HJ4). The Bureau for smaller animals, such as rodents. In He also noted that the test compound relies upon: (1)the fact that Dr.Hsu human beings,and often in larger was tested for impurities and that none describes breaks and gaps without animals, blood cells are used instead were found (id. at 26).However. the ALJ distinguishing between them in terms of which are obtained by simple,well emphasized that two other investigators biologicalsignificance(A-aooat 4): and known procedures. (Ford.A-.219 and Dick. A-l77) failed to (2)Dr.Legator'stestimony that certain Once the cell specimens are obtained replicate Legator's results, despite other, unnamed scientists recently and prepared, they are examined "appear(ing) to have used the exact stated that gaps were as significantas microscopicallyfor the type and protocol used by Dr. Legator."(id. at 25). breaks (G-124 at 15J. The Bureauadmits, frequency of chromosomalaberrations. The ALJconcluded,therefore, that"the however, that gaps may indeed be H the results from the test group are Inability of replicating (Legator's) results entitled to less weight than breaks ''positive,'' they are compared to the puts themin doubt," (id. at 26). negative control for statistical Analysis: '(Bureau's Briefat In (3) Abbott takes no 10:HJ4). response, significance.In contrast. if the results exception to this aspect of the ALfs Abbott contends that gaps are entitled from the test group are "negative," the to no weight at all because: (a) they may decision.The Bureau's position on the results are compared to the positive replicability issue is that "the Dick study be quicklyrepaired by cellular control to ensure that the experimental was not an exact duplication and has mechanisms;(b) they are difficultto environmentwas conducive to obtaining problems of its own" (Bureau'sBriefat detect; and (c)in any event, are not a positive response. 77). The Bureau also suggestspossible heritable (Abbott's Exceptions at 36; 4. Suggestive Studies. As noted above, bias in the Dickstudy becanse Dr. Dick Abbott's Briefat 48-52). the administrative record contains six was an Abbott employee at the time the As seen from the definitionsof breaks studies whose findings are inconclusive study was performed (id. al76-77: 87). and gaps, the differencebetween the but suggestiveof mutagenicity.Each The Bureauhas made no commentson two is'one of degreerather than kind study's findingswere statistically the Ford study. Le; breaks are wider separations than significantat the P<;.05level. The reason As to the quality of the Legatorstudy are gaps, but both are separa~ons these studies are "suggestive"rather itself, Abbott argues: (a) thatLegator's nonetheless. It logicallyfollows that than "positive" is that the findingswere findings of breaks do not constitute . gaps, like breaks, are entitled to some primarily of breaks rather than ''permanent'' breaks (Abbott's Briefat weight,for gaps may develop into exchange figures(see discussion in 56-57); and (b) that Legator did not use breaks. Subsection C.l.b. above). positive controls (id. at 57).The Bureau's However, as Dr.Legatorhimselffound response is tliat breaks db constitute in his collaborativein vivo cytogenetic a. Legator, et al; (G-B) (bone marrow Significant genetic events (Bureau'sBrief study (A-239) described in Subsection portion) at 101H)4); and that positive controls are F.2.a.above, "•.• gaps are the least (1)Study Design: This study was not necessary to validate positive conclusivecriterion in determining performed on Holtzman strain albino results, only negative ones (id.'at 76). cytogeneticeffects.The variability male rats usingCHAas the test Viewed by itself,I would characterize between laboratories was greatest for compound.Five test groups of 20 to 3{) the Legatorstudy as a VelY well gaps •••" (A-239 at 350). Based upon rats each were formed,and each group designed experiment which produced this conclusion, I find that gaps are was given daily CHAintraperitoneal clear positive findingsof breaks. By the entitled to considerably less weight than iDjections of 1, 10,20,40 or 50mg/kg, phrase "very well-designed."I mean are breaks. I would therefore classify respectively, for a period of five days. A that Dr.Legator employed a sufficient gaps in terms of "additional support" similar-sizednegative control groupwas number of test animals (2G-30 per dose rather than as "primary evidence" of also established and given dally group)and analyzed a sufficientnumber potential mutagenicity. ~ injections of disUlledwater over the of cells (625 per dose group). 3. Conduct ofan In Vivo Cytogenetic same period of time.The animals were Additionally, as the ALJnoted, he tested Study. JI.,n in vivo cytogenetic'study is sacrificed 24hours after the last the CHAfor impurities and none were carried out in three principal steps: (1) injection,and slides were prepared and found.He also coded the slides to dosing; (2)obtaining and preparing cell analyzed for 625cells at each dose level, prevent possible bias. Finally, he used a ... 61514 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1900 I Notices dose range (five doselevels] which 47623). The parties have since spermatogonial cells carry special permitted dose-response information tq stipulated, however, that the requested significance from a mutagenicity be obtained. By the phrase "clearly data could not be located (Stipulation standpoint While positive findings In positive fmdings," I refer both to the dated September 17,1979 at 8). Although somatic cells (e.g., bone marrow cells statistically significant findings at four I agree with the former Commissioner and blood cells) help us learn whether a out of five dose levels, and to the that the requeste(information would certain compound causes chromosomal consistent dose-response trend which have been helpful, I find that I can aberrations at all, positive findings In was observed. 11J.e dose-response, adequately evaluate this study on the spermatogonial cells give us the added especially, adds credence to the positive basis of information currently in the ' information that those chromosomal findings (G-124 at17; see G-18 at 2). record. aberrations occur in the very cells that I have already considered and b. Legator, et al. {G-B} [spermatogonial determine heredity (G-124 at 16-17). dismissed Abbott's first contention, o Thus, this study presents Important regarding the biological significance of cellportion) , evidence that genefie abnormalties breaks (see Subsection F.2.c. above.] (1) Study Design: The design of the caused by cyclamate or its metabolites Abbott's second criticism, regarding the spermatogonial cell portion ofthis study may be passed on to future generations. lack of positive controls, is also without was identical to the bone marrow With respect to the Friedman study merIt. I agree with the Bureau that portion discussed above, except that 500 (A-195),I agree with the ALJthat this positive controls are not always . metaphases were analyzed for each . study is distinguishable from Legator's necessary to validate positive findings dose level (rather than 625)~ because of the difference in test (Tr. at 975; see discussion in Subsection . (2)Study Results: The investigators compounds and routes of C.3.c.above), especially where, as here, observed a statistically significant administration. Moreover, for the' the testing laboratory has historical increase (P<;;.05) over controls in the additional reasons stated in Subsection control data (Tr. at 960). percentage of cells with breaks in each F.6.a.(4j below, I find that this study is Furthermore, I disagree with,the ALfs of the five dose groups. The- deficient and therefore entitled to no finding that "the inability ofreplicating investigators also found a linear dose- weight at all. ' [Legator's] results puts them in doubt" response throughout these five groups. With respect to the Ford study (A- (id. at 26).I disagree because the two Finally, as with the bone marrow 297), however, I strongly disagree with allegedly "replicate" studies (Ford, A portion. the authors noted "infrequent the ALJthat is was a replicate of 297and Dick, A-177) are each exchange figures" whiChpresumably did Legator's study. Aslexplained in mi' "deficient." Ford administered CHA to . .not reach statistical significance. analysis of the bone marrow portions of only one group of three animals and The ALJ characterized this portion of these two studies (the design of each then analyzed a total of only 150 the Legator study as showing an ' investigator's sperm cell portion being metaphases. This test population is "adverse effect" (id at 35), and, more virtually identical to the design of his simply too small for any weight to be specifically, as demonstrating a own bone marrow portion), Ford simply given to it's "negative" results. It statistically significant increase in used too small a test population (1 dose certainly in no way detracts from "breaks" in the treated aniinals over the, level; 3 animals; 124 metaphases) for it Legator's positive findings which were controls which was found to be dose- effectively to rebut Legator's findings , derived from an experiment employing' related (id at 27-28).The ALf also noted • (which were based on 5 dose.levels; five groups of 20-30 animals per 'group that "infrequent exchange figures" were over 100 animals, and 2500total with the total number of metaphases observed (id' at·28). The ALI metaphases]. Moreover, Ford's study examined exceeding 3000. distinguished the Friedmanstudy (A- size is so small that rhave classified it The Dick study (A-l77) is deficient for' 195)because the Investigator-there used as "deficient" and have attributed no a different reason. Although it is true a different test compound (cyclamate weight to itatall (see Subsections F.O.a•• that Dick reported no statistically . rather than CHA)' and a different and d. below). significant increase of breaks in the method of administering that test (4) Other Matters: My earlier treated over the negative control group, compound (feeding rather than comments (in the bone marrow section) neither did Dick find such an increase of intraperitoneal injection) (id at 28). regarding the former Commissioner's breaks in the positive control, However, the ALI did note that the-Ford request for additional data are equally triethylenemelamine (TEM), as _ . study (A:'2971"appe~[ed]to have-used applicable here. compared to the negative control (A-177 the same protocol as Dr. Legator but . at Table 11I). This absence of breaks in obtained negative results" (id at 28). c. Majumdar andSolomon.{G-26} the positive control demonstrates that (3)Analysis: Abbott agrees with the (1)Study Design: This study was Dick's study was so insensitive that she ALJthat Legator's findings should be performed on Mongolian gerblls using could not-even find breaks in a discounted because they could not.be. calcium cyclamate as the test • compound known to be mutagenic. A . replicated by Ford (Abbott's Exceptions compound. The design of the study was fortiori, no' conclusion can be drawn at 65).Other than that, Abbott has not . similar to that of Legator, et al, (G-9) in from Dick's failure to observe' breaks in raised any additional criticisms.directed that the test population consisted of five the test compound, CHA.~6 towards the design ofthe dose groups (ten animals per group) that (4) Other Matters: Former spermatogonial cell portion of this were given daily injections of 10, 30,50, Commissioner Kennedy, in his Remand study. Moreover; Abbott explicitly does 70 or 100mg/kg. respectively, for a Order, requested the parties to provide ' not take exception to the manner in period oHive days. A negative control .certain underlying data to this study which the ALI distinguished the group 'often animals received injections because it would be "helpful" in Friedman study (see id. at 65)~ The of distilled water over the same period evaluating the study "more fully" (44 FR Bureau makes no additional comments of time. All animals were sacrificed on on these Issues. the fifth day. Cells selected for nnalyills 20 1do note. however. that. contrary to the I would adopt here, by reference, my were from the bone-marrow and suggestion raised by the Bureau. the mere fact that earlier evaluation ofthe design and numbered 3~350 per dose group. Dr. Dick was an Abbott employee at the time her study was conducted Is. by Itself. simply not results of Legator's bone marrow (2)Study Results: In the four highest relevant to the issue of Investigator bias. portion. Moreover, positive findings in dose level groups, the investigators Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61515 reported a statistically significant and chromosome aberrations" (id. at 27, a mutational agent which causes increase (p<.001) over controls in the' 35). chromosomalaberrations (Tr.at 718). percentage of cells with breaks as well (3) Analysis: Abbott takes exception second, contrary to Abbott's assertions, as in the percentage of cells with gaps to the low number of animals utilized in the authors do not concludethat the and fragments. Moreover, across these the study, and, relyingon alleged breakage was not dose-related.In fact. four dose levels, a slightbut consistent conclusionsby the authors, claims that: the authors concludedthat they dose-responsewas found. (a) CHAis a "clastogen" only;(b) observed a "dose-effectcorrelationin The ALJ characterized this study as breakage was not dose-related:and (cl the frequencies ofboth major and minor having "positive results" due to the CHAdoes not induce translocations aberrations •••" [A-725 at 411; G-44 at findingswhich I have just summarized (Abbott's Exceptionsat 60). The Bureau 411}. This would include both exchange lid. at 35). responds only to the issue of study size, figuresand breaks, respectively.Finally, (3) Analysis: Abbott makes no specific arguing that where positive results are Abbott's counsel apparently misread exceptions to this study. I agree with the found in a small study, this is an page 410of this study. Nowhere on that ALJ that this study produced clear, indication that the test compoundis page does the word "translocations" statistically significantpositivefindings, quite potent (Bureau'sReplyat 22). appear. although the word especiallyin lightof the dose-response The major strength of this study is the "tranformation" does. This exception. obtained (see G-124 at 17). Moreover, statistically significantfindings of major therefore,requires no response. the investigators employeda sufficient structural aberrations, which include number of animals (ten per dose group) exchangefigures. As noted previously, e. l'an Went-de l'nes (G-45) and analyzed a sufficientnumber of these are the types of chromosomal (1)StudyDesign: In this study, cells (300 per dose group) to give added abnormalities which the parties agree Chinesehamsters were given CHA credence to the results. These findings cause heritable genetic damage throughoral intubation (forcedfeeding). tend to confirm the results of the bone (Abbott's Briefat 49): Bureau's Briefat Twenty hamsters were each dosed with marrow portion ofLegator, et al, (G-9) 100). Moreover, because a lamb is a 200ms/kg for three successive days. (see G-121 at 13), and thereforeenhance muchlarger animal than the The cells analyzed were peripheral the credibilityofboth studies. Although conventionallyused rodent, these blood cells.Bloodwas drawn both the number of chromosomal aberrations findingshave a more direct appUcability before and at the conclusionof the dose observed in this study was somewhat to man (Tr.at 9(4). Finally,the period: thus, each animal served as its lower than that reported by Legator, et demonstration of a dose-response,as own negative control.A total of1,000 aI. (G-9), this couldbe explained either noted in the previous studies greatly metaphases were analyzed for both by the fact that Majumdarana Solomon enhances the credibilityof the positive treated and controlgroups.The slides to used calciumcyclamaterather than its results (G-124 at 17). Once again, this be analyzed were coded so that the metabolite,CHA(G-124 at 20), or by the analysis is supported by the relevant persons reading themhad no knowledge fact that they used a differentanimal expert testimony(id. at 17-18). of whether they came fromtreated or species (see discussionin Subsection I have reviewed in general terms control animals. D.2. above).In any event, this study Abbott's criticismthat this study (2) StudyResults: The investigator stands virtuallyunimpeached. This employedtoo few animals (see found a statistically significantincrease evaluation is further supported by discussion in Subsection C.3.a. above), [p<.005} over controlsin the total expert testimony (G-124 at 20). but I will elaborate here. It is true that number of structural chromosome Turner and Hutchinson employedonly abnormalities.The findingsincluded do Turner andHutchinson (G-44) one animal per dose groupper treatment several exchangefigures, one ring. and (1) Study Design: This study was period. It is also true that this is far less numerousbreaks and fragments. carried out on fetal lambs usingCHAall than the ideal number of animals to use. The ALI characterized this study as the test compound. Each treated animal For example,had thisstudy produced "positive" with an "increase in received one dose of CHAof either 50, negtiveresults, it couldhave been justly structural aberrations" (id. at 26, 35). 100. 200, or 250 mg/kg. The animals were criticizedfor being too insentitive (3) Analysis:Abbott criticizesthis giventhe CHAin utero by intravenous because there would not have been a study in three ways: (a) absence of injectionsover a period of either five or sufficientlikelihoodof detecting . positive controls: (b) difficultyin 18hours.Eighttreated animals were mutageniceffects.even if present. On evaluatingfindingssince all types of used in all, one for each dose level and the other hand. where, as here, a study aberrations were grouped together;and dose period.In addition. one control with few animals produces positive (c)allegedlysmallincrease in total animal was used for each dose period. results, it cannot be critized for being aberrations in treated oyer controls Cells obtained for analysis were too insensitive. Quite the contrary, what (Abbott's Briefat 58-59). In response. peripheral blood cells that were drawn such a test suggestsis that the test the Bureausimplyemphasizesthe fromeach fetus. Resultswere based on compoundis sufficientlypotent that it is positive findings(whichincluded 'the analysis of a combinedtotal of500 capable of being detected by even an exchange figures] and the extra controls cells. insensitive test (Tr.at 941). Therefore, employedby the investigator,such as {2) Study Results: The investigators althoughTurner and Hutchinson's the precautions taken to ensure purity of reported a statistically significant findings would have been strongerif the test compound(Bureau'sBrieF-at 75 increase over the control (in each time their test population size had been 75). period) in three differentcategories:(a) larger, the findings based upon the The strength of this study lies in the percentage of cells with major structural populationused are nevertheless valld, statistically significantincrease over aberrations [i.e.,exchangefigures): (b) Abbott's other exceptions regarding controls in terms of tota! chromosomal percentage of cells with breaks; and (c) the author's alleged conclusionsare aberrations found,and the fact that this percentage of cells with total totally without merit.Althoughthe included exchange figures(see Tr. at aberrations. In addition. a linear dose authors do conclude that CHAmay be a 9Z1). Unfortunately, due to the design of response was found. "clastogen," this in no way advances the study which called for only one dose The ALJ characterized this study as a Abbott's cause.The company's own level. no dose-responseinformation "positive" one, showingboth "chromatid witness, Dr.Hsu, defined a clastogen as could be obtained. 61516 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices There is some merit to Abbott's individuals on cyclamate) over the' aimed at maximizing protection of the criticism regarding the small increase-in Control I group (diseased individuals on public health, I am interpreting the the amount of total chromosomal placebo) in terms of: (a) percentage of positive results of this study as having aberrations in treated over controls. cells with chromosomal aberrations (i.e.. been caused by the cyclamate (i.e., tho Although the findings were statistically breaks, gaps and exchange figures first option just discussed). This significant at a highconfidence level grouped together) (p=.032): (b) total interpretation is at least as likely to bo (P<.0Q,5), the actual number of number of breaks (p=:.05):and (cl correct as the synergistic effect . aberrations observed was relatively low aberrations with.open breaks (p=.038). interpretation, and Abbott has not (see G-45 at table 2). This reduces the No significant difference was found for satisfactorily shown the contrary to bo biological significance of the findings. the fourth category labeled true. Indeed, this study presents a perfect "restructurings," which would Include Abbott's remaining exceptions can bo example of how the concept of exchange figures. dealt with.briefly. First, Abbott biological significance can reduce the The investigators also found, challenges the small population size of weight otherwise accorded to however, a statistically significant 1 statistically significantresults. While increase in the Control I group (diseased .fhe treated (11persons} and Contro I this does not mean the findings of this individuals on placebolover the Control (10persons) groups (Abbott's study should be completely discounted, II group (healthyIndividuals onplacebo) Exceptions at 62),as well as the the weight given is not as great as in the in terms of: (a) total number of breaks disparity in size between those two previous studies discussed. _(p=,024): and (b) restructurings (P<:.OO3). groups and Control II (52persons) (id.). Abbott's complaint regarding the No significant difference was found in Abbott also questions the validity of . ,grouping of the chromosomal the other two categories. analyzing only about half as many calls aberrations has less merit. Although the (3)Analys4: The strong points of this \ per person in Control II (55) as in the findings would have been somewhat study are that: (a) it was conducted on other two groups (100) (id.). However, stronger if exchange figures had been humans: (b) it used dose levels "that are there is unrebutted testimony that the grouped separately and found to be frequently encountered in individuals size of the testpopulation was,adequate statistically significant (as was true with who are consuming cyclamate" (G-124' (Tl'.at 508)and that an ample number of the )'urner and Hutchinson study), so, at 22):and (c) theinvestigators found a cells was analyzed (Tr. at 515).I agree too, would the findings have been statistically significant increase in . with this testimony, and I would again somewhat weaker ifno exchange figures breaks in the treated over the Control I emphasize that the actual size of a test had been found at all. Thus, the findings I group. Dr. Legator termed this study population is less important where , simply are what they are. - '''probablythe most relevant piece of positive fmdings are obtained. Finally, Abbott's third comment information we have right now to be Abbott next seeks to clarify the exact regarding the absence of a positive expanded on" (Tr. at 974).The ALJ nature of the positive flndlngs by control is without merit. As noted above apparently agreed (see extended stating: "the only chromosome in Subsection C.3.c.above, positive discussion in at 27:35).(See also G-121 aberrations that were significantly controls are not necessary to confirm at 13: G-123 at 4: G-124 at 21-22.) increased were 'open breaks," (Abbott's positive findings (see TR. at 975). Although Abbott takes numerous Exceptions at 61).Abbott is generally f. Bauchinger, et al. (J-l): (1) Study exceptions to the ALI's findings on this correct on this point. For a more precise Design: This was the only suggestive study. only one of these requires a statement of Bauchlnger's findin~81 tile study conducted on human beings. detailed discussion. This relates to the . my statement of the Study Results, 'Cyclamate was the test compound. The possibility that a synergiStic effect could supra. Abbott also states that "similar , treatment group consisted of 11 persons, nave beenatworkbetween the kinds of chromosomal aberrations and all of whom suffered from liver or cYclamate and the diseases. In.this frequencies were observed in both the kidney disease(s). Each was fed either 2. regard, there are two ways inwhich to treated group and control group I" granis or 5 grams of cyclamate per day . interpret BauChinger'sfindings. The first (Abbott's Exceptions at 63).Abbott i6 .' for periods ranging from 310 to 1160 is to say that sinec the cyclamate was also correct here in so far as days. In addition, two control groups the only differing factor between the . "chromosomal aberrations" refers to were established. The first ("Control I") treated- and the.Control I group, it was Bauchinger's "Restructurings" category, consisted of 10 persons with the same or the cyclamate which caused the ch ul I d ch fi similar diseases as the treated group. _ increased incidence of breaks. This view whi wo d inc u e ex ange igures Control I was given fructose instead of _ was adopted by the ALJ (id at 27).The U-l at Table 3).Abbott's point, I believe, cyclamate. The second control group second possible interpretationis that the is that the only statistically significant ("Control II") consisted of 52 healthy increased incidence of breaks was. findings were in terms of breaks and not exchange figures. This is true, as I have persons. The au thors ma de speci al causedby a synergistic e ffiect, i.e., the already pointed out. To the extent that mention of the fact that none of the combination of the cyclamate and the ' participants in this experiment received diseases. This theory is somewhat the ALI's decision suggests anything to . therapeutic radiation or thereapy with supported by the increased incidence of the contrary, I would modify it alkylating drugs. The blood cells used .J>' breaks found in the diseased control accordingly. for analysis were peripheral group over the healthy control group. Given these findings of breaks, Abbott lymphocytes. Approximately 100 cells However. even were this second criticizes their validity because they were analyzed from each individual' interpretation the proper one, the results. were not related to dose or duration of from the treated and Control I groups, of the Bauchinger study would still be exposure (Abbott's Exceptions at 61). and 55 cells from each member of the "relevantbecause, if approved, However, Dr. Legator testified that ha Control ngroup. cyclamatewould be.ingested by a broad would not expect to find a dose (2) Study Results: The results of this segment of the population, including response relationship in a human study study are most-clearly presented in those with kidney andlor liver disease" of this size (Tr. at971-?2).I agree with Table 3 U-1).Here the investigators (id. at 35)., ,theBureau's position on this point. In reported a statistically significant . Consistentwith-the cautious approach reaching this conclusion, I note.that increase in the treated group (diseased taken throughout this decision, which is Abbott has not produced any expert Federal Register I Vol. 45. No. 181 I Tuesday, September 16. 1980 I ?':otices -51517 testimony to lend scientific credence to regimens of distilled water. No unsubstcntlated, especially in li.;ht of its theory. concurrent positive controls were used, van We:l~·de Vries' pcsitiva f:u:dings Finally. Abbott raises two concerns After sacrifice. cells were obtained from with Ch:nese hamsters (G-45J). regarding possible confounding the bone marrow for analysis. Therefore. consistent 'with the approach variables. FIrst. Abbott questions the (2) Study Results: The authors outlined in Subsection C.3.e. above, I validity of using patients in the treated reported no statistically significant flnd that Brewen's lack of a positive and Control I groups with "similar" increase in the treated animals over control reduces the study's weight but rather than "identical" diseases controls in terms of chromosome does not undermine its overall validity. (Abbott's Exceptions at 61). The aberrations. These findings were based Finally, I consider it important to note Bureau's expert testimony. however. on analyses of either 200 or 400 cells per that in the highest dose group (1350mg[ dismisses this-Abbott concern (Tr. at treatment group. The authors did note. kg lotal dose), half of the animals died 970-71),and I agree with this unrebutted however. that half of the treated animals before the end of the experiment. I evidence. Second, Abbott claims that at the highest dose level died before interpret this to mean that cells from the patients' exposure to diagnostic completion of the experiment. these animals were not examined for radiation and to non-alkylating drugs The ALI found that this study mutagenicity. This inference is confounded the study's results (Abbott's "evidenced no chromosomal damage" supported by the fact that 0n130' half as Exceptions at 62).and that the ALI's (id. at Z6). many cells (200)were examinedin this finding that cyclamate was the (3)Ana/J'sis: Abbolltakes no dose group as compared to the middle "causative factor" of chromosome exception to the ALI's finding (Abbott's dose group (400cells].I therefore damage (breaks) (ID at 27)rested upon Rxceptions at 57). and asserts that the consider the results from this one dose assumptions unsupported by the record study is important because it used a level to be invalid because mutagenic (Abbott's Exceptions at 63).A review of dose level five times that used by effects could have been masked by the the record. however. has shown there Legator (Abbott's Brief at 57).The fatality of the dose. also to be unrebutted expert testimony Bureau criticizes the study because: (a) I also note that having eliminated this in the Bureau's favor on this issue of the size of the treated groups was not upper dose level. Brewen's highest valid confounding variables (Tr. at 526),and I specified: and (b) no positive controls dose level (450 mg/kg total dose), agree. The very purpose of using the were used (Bureau's Brief at 85-86). The although greater than Legator's (150mgt Control I group (having similar diseases Bureau stressed that positive controls kg).was well below van Went-de Vries' as the treated group and therefore are especially necessary where, as here. (600mg/kg) Dighest total dose which similar exposure to diagnostic radiation a test animal whose sensitivities are not also involved CHA. Thus. Brewen's and non-alkylating drugs) undoubtedly _well known is used (id. at 86). findings do not pre-empt the suggestive was to eliminate the very kind of I agree with the ALI and with Abbott studies in terms of dose size. confounding variables which Abbott is that Brewen, et al, did not find a In conclusion, the Brewen study raising. Moreover. the ALI's conclusion statistically significant increase in presents inconclusive evidence that that cyclamate was the "causative chromosome aberrations. The study is CHA does not cause chromosomal factor" of chromosome breaks is more therefore "negative" in the general aberrations in the bone marrow of than adequately supported by the record sense. However, I also agree with the Chinese hamsters. I would have more (G-121 at 13; G-123 at 4; G-124 at 21-22). Bureau that the study has shortcomings confidence in these results ifBrewen I therefore conclude that the which reduce the weight to be accorded had specified an adequate test Bauchinger study presents statistically to it. population size, and ifhe had presented significant findings of breaks which are First, the authors' failure to specify the adequate positive control data. strongly suggestive of cyclamate's number of animals used raises a b. Cattanach. et al. (A-151). (1)Study mutagenic potential. question which is not answered by the Design: This study was conducted on 5. Negative Studies: The current record. Although I might be mice using CHA as the test compound. administrative record also contains four justified in rejecting this study Different sized lest groups were used. in vivo cytogenetic studies which I have altogether as "deficient" due to this The first group of four mice were given classified as "negative"-i.e., the studies shortcoming (since it is Abbott's burden daily CHA injections of 50 mg/kg body meet the minimum criteria set forth in to establish its proof to a reasonable weight for five days. The second group Subsections C.2. and 3. above and found certainty), the fact that 400 cells were consisted of eight mice which received no statistically significant increase over analyzed at the middle dose level daily CHA injections of 100 mg/kg body controls in the types of chromosome suggests that at least at that dose level weight, also for five days, The negative aberrations which were scored. These the study size may have been sufficient. control group. consisting of eight mice, studies were obtained from the bone (Note, for example. that in the received distilled water. No concurrent marrow of Chinese hamsters (A-143), Majumdar and Solomon study. G-:6. 10 positive controls were used. The cells the spermatocytes of mice (A-151), and animals and 300-350 cells were used per examined were spermatocytes and the spermatogonia of Chinese hamsters treatment group.) Thus. I consider this numbered 200per animal (which total (A-716 and A-811, App. 19). As unknown fact to affect the weight but 800for the first group and1600for the explained in Subsection F.l. above. not the overall validity of this study. second and control groups). On130T however, the findings are insufficient to Second, Brewen's failure to use a translocations (i.e.• exchange figures) outweigh the suggestive in vivo positive control also reduces the study's were scored. cytogenetic findings just described. weight, While it is true that Brewen may (2) Study Results: The investigators a. Brewen, et al: (A-143). (1)Study have had adequate historical control reported no statistically significant Design:This study was performed on data. none was presented, and. again. it increase of translocations in either Chinese hamsters using CHA as the test is Abbott's burden to ensure this treated group over controls. Infact. no compound. Three test groups of information is presented. I also note. translocations at all were observedin unspecified size were injected daily for however, that no other evidence exists the treated groups. although one was three consecutive days with 50, 1:;0or suggesting that this study was seen in the control group. 450mg/'kg bodyweight. Negative insensitive (the Bureau's claim that the The ALI found this study to have control animals were given identical test animal is insensitive being produced "negative" results (id. at 28). 61518 Federal Register I Vol. 45, No.'18i I Tuesday, September 16",1980 / Notices (3)Analysis: Neither party has taken data were available to validate the Specifically, Dr. Green testified that exceptions or commented in any detail negative findings. cyclophosphamide dose levels employed on this study. The purpose of the study c. Larke, et al. (A-716). (1]Study by Lorke normally produce was to determine whether CRA induces Design: This study was conducted on chromosomal aberrations in 20% to 40% exchange figures in mice. This type of Chinese hamsters using CHA as the test of cells examined, whereas Dr. Lorke's study is important in evaluating the · compound- The CHA was administered . values did not exceed 9%(II'r. at 851).1 mutagenicity of a compound because/as · orally to one group of eight hamsters at find the Bureau's testimony persuasive the parties themselves agree, exchange a dose of approximately 100 mg/kg body on this point, especially since Abbott figures are one means by which genetic weight/day for five consecutive days. produced no expert testimony to the abnormalities can be transmitted to Negative and positive control groups of contrary. Arguments by Abbott counsel future generations (Abbott's Brief at 49; ,eight animals each were run miss the point for two reasons, First, Bureau's Brief at 100). One of the concurrently. The negative controls even if the positive control values ore Bureau's hypotheses is that breaks are were not dosed. The positive controls statistically significant, those values con significant because they may lead to were given cyclophosphamide at a dose still be lower than-would be anticipated. , exchange figures. (Bureau's Brief at 101; of 100 mg/kg body weight/day for five Second, even if.Dr. Lorke validated tho see general discussion inBubsection days. 100 spermatogonial'cells from sensitivity of the Chinese hamster in F.2.c.(1) above). The Cattanach study, in each animal were then analyzed. earlier tests, it is still quite possible that essence, is designed to test that (2) Study Results: The investigators something in the 'current study reduced hypothesis with respect to cyclamate. scored the cells for three types of the sensitivity of the results at issue. I I agree with the ALJ that this study is , chromosomal aberrations: (a) cells therefore attribute less weight to this negative, but I find that it has two containing aberrations;including gaps; study than would Abbott. internal shortcomings which must be (b) cells contalnlng aberrations, without I do not, however, find merit in the noted. The first shortcoming is that the gaps; and (c) cells with translocations. . Bureau's criticism of Dr. Lorke's lower dose treatment group contained No statistical difference between the statistical analysis. As Abbott correctly only four mice, and these findings must treated and negative controls was found points out, the Bureau did not present therefore be rejected due to that group's in any of these categories. In fact, no . any evidence demonstrating that Dr. insensitivity. (See discussion in translocations were observed in either Lorke's findings would not have reached Subsection C.3.a. above.] This of these two groups. A statistically statistical significance if another method shortcoming, however, does not significant increase in all categories was had been used. Moreover, the Bureau's substantially reduce the weight given to observedin the positive controls when witness on this point, Dr. Green, is not the study as a whole because the compared to the negative controls. himself a statistician and instead based findings of the higher dose group are The ALJfound this study to produce his testimony on his conversations with valid. That group contained eight "negative" results (id. at 28). other, unnamed persons (Tr. at 853)who animals. This number is sufficiently (3) Analysis: The Bureau criticizes this were not available for cross' close to the guideline to ten which I study on two principal grounds: (a) that examination. have followed (see Subsection C.3.a. the positive control values were I therefore conclude that this study and G-124 at 18] to be considered unusually low, suggesting an presents inconclusive evidence that adequate, especially in light of the large insensitivity in the test (Bureau's Brief at CHA does not produce chromosomal number of cells (1600] analyzed. , 86-87; Bureau's Reply at 26):and (b) that aberrations in the spermatogonia of The second shortcoming is the lack of. Dr. Lorke employed an inappropriate Chinese hamsters. I would have more 'a positive control. For the same reasons method for statistical analysis (Tr. at confidence in these results if the discussed immediately above in 853).Abbott counters the Bureau's first positive control-data had been within connection with the Brewen study, I find argument by stating that the sensitivity the normal range for that compound. that this shortcoming reduces t}1e weight of the Chinese hamster spermatognia d. Larke, et 01. (A-811, App. 19). (1) but not the overall validity of were validated by earlier studies, and Study Design: This study was also Cattanach's findings. that the statistically significant positive conducted on Chinese hamsters but I therefore fmd that this study control values validated the sensitivity used sodium cyclamate [rather than presents inconclusive evidence that of this particular study (Abbott's Brief at CHA) as the test compound. Six CRA does not induce exchange figures 62;Abbott's Exceptions at 66).As to the hamsters were orally given 2,000mg/kg in mice. I emphasize the appropriateness'of Dr. Lorke's statistical body weight/day of sodium cyclamate inconclusiveness of these findings method, Abbott contends: (a) the chi for five days. A negative control group because exchange figures are rare square method is appropriate: (b) the of six hamsters were not dosed. Two events that are difficult to detect (G-124 only testimony elicited by the Bureau is sets of positive controls were also run at 30;Tr. at 956).As Dr. Cattanach based on hearsay: and (c) the Bureau concurrently. In the first, six hamsters. himself admitted: did not show that the results of the received 1,000mg/kg body weight/day study would be any different ifany of trimethylphosphate orally for five ••• but here a word of caution must be introduced. For the induction of statistical test had been used (Abbott's days. In the second, six other hamsters translocations at legst two breaks must occur Brief at 62-63). , received 250mg/kg body weight/day of in the same cell at the same time and the I agree with.the ALJ and with Abbott cyclophosphamide orally, also for five broken chromosomes must rejoin in such a that this is a "negative" study, but I also days. 600 spermatogonial cells (100per way that each rearranged chromosome agree with the Bureau that the postive hamster) were analyzed in both the possesses I centromere. A Failureto detect control values raise a question about the sodium cyclamate group and negative translocations is not thereForeincompatible study's sensitivity that reduces the control group. The total number of cells with spermatogonial chromosome breakage. weight I would otherwise have given to work~s analyzed varied for the positive It is clear that more needed. it. The Bureau adduced testimony from controls. (A-151 at 474).Given'the apparent two expert witnesses that the positive (2)Study Results: The cells wore difficulty in detecting exchange figures, I · control values for cyclophosphamide scored in the same three categories would have more confidence in the reported by Lorke were 'well below the described above in connection with tho results of this study if positive control ' norm (Tr. at 851;G-124 at 19-20). other Lorke study (A-716). The , Federal Register I Vol. ,45, No. 181 I Tuesday, September 16, 1980 I Notices 61519 investigators found no statistically type of cells analyzed: bone marrow, The cells numbered 700, 850, 600,470 significantincreasein any category ill blood, or sperm cells. and 400, respectively, for thefive groups. the treated group-over the negative a. Bone MarrowStudies: (1) Collin (b) StudJI Results: Cells were scored controls. Analyses of the positive (G-27). for two categories ofchromosomal control groups yielded statistically (a) Study Design: This Was a rat aberrations: (i) gaps and breaks significantincreases in all categories feeding study using sodium cyclamate (combined); and (ii) reunionfigures and when compared to the negative controls. as the test compound. The test fragmented metaphases (combined). The The ALJ found these results to be population consisted offour rats. The investigators reported fewer gaps and "negative" (id. at 28). dose size was stated in terms ofbeing breaks in the treated gronps thanin the (3)Analysis: The parties apparently 5% of the feed. The length ofexposure negative control. No reunion.fignres or directed their commentsmade in ranged from two to six months. The fragmented metaphases were found in connection with the other Larke study number of cells analyzedwas not any of these three groups. For the (A-716) to this study as well. In specified. There is also no mention ofa positive controls, there was a addition, the Bureauaiticizes this negative control. statistically significant increase for particular.studyfor only using six (b) Study Results: The investigator METEPA over negative controls in both animals per group (Bureau's Brief at 86 reported chromosomal damage, categories. ForTEM,however. there 87). including breaks. butno data were was a statistically significantincrease I adopt herebyreference my earlier presented and no analysis ofstatistical only for the second category.Forgaps .analysis of the other Lorke study (A significance wasreported, and breaks, the findings forTEMwere 716) with respect to the positive control The ALI characterized this study as virtually the same as for the negative and statistical methodology Issues.' . "positive" (ia. at 35), based upon results control (A-177 at TableTII). In addition, 1agree with the Bureau which included "chromosome breaks, The ALI found thisstudy to be that the number ofanimals per group in the absence of satellites on negative despite "appear[mg]10 have this study (six) is too small. Under the chromosomes and numerous achromatic used the exact protocol usedbyDr. criteria set forth in Subsection C.3.a. areas" (ia. at 26). Legator" (id. at 25; see alsoid. at 35). above. I wouldbe justified in totally (c) AnaIysis:Abbott takes exception Thus, the ALJ concluded thatDr. Dick's rejecting this study as deficient because to the ALI's statement that Collin found findings helped place Dr. Legator's in its sensitivityis too low. However, this "numerous achromatic areas," claiming doubt (id.). low sensitivityis partially offsetby the that this finding came from the in vitro. (3)AnalJ'sis: Abbott agrees with the extremely high dose used. 2.000 mg/kg not in vivo,'portlon ofthe study ALJ (Abbott's Exceptions at72; see also body weight!dayfor five days. (Abbott's Exceptions at 57). Abbott also Abbott's Briefat 56-57). The Bureau (Compare, for example. Majumdar and criticizes the study's design for.using too criticizes the Dick study on two related Solomon (G-26) which used doses of few animals and analyzing too few cells grounds.FlISt. the Bureau challenges the calciumcyclamate on Mongolian gerbils (id.). The Bureau does not discuss this validity ofDiel's grouping of"reunion of 100 mg!kg body weightlday for five study in any detail, butmerely lists it as figures" and "fragmentedmetaphases" days.) one of a group that "produced clear together, arguing that the former are the I therefore conclude that this study by positive results" (Bureau's Brief at 75). best indicators of heritable genetic Lorke, et al. presents inconclusive Unlike the ather studies with positive damage while the latter are theleast evidence that sodium cyclamate does findings discussed above, Collin's work reliable [Bureau's Briefat 89). Assuming not induce chromosomal aberrations in is deficient and warrants no weight at this to be true, the Bureau attempts to the spermatogonia of Chinese hamsters. all. The primary deficiency in this study eliminate thefragmented metaphases I would have more confldence in these is that it contains no meaningful from the incidence found for the positive negative results ifthe positive control presentation of data, and therefore control. TEM, and then asserts that the values for cyclophosphamide had been insufficient information exists to remaining incidence for reunionfigures in the normal range and ifDr. Larke had evaluate it properly. I also nate that the for TEMis far 100 low (id. at 89-90). used more animals per group. study contains no comparison, in terms I agree with the ALJ 10 the extent that (4) OtherMatters: For the record, I of statisticalsignificance, between the Dr. Dick used a very similar protocol as note that the sodium cyclamate portion findings of the treated group and a Legator, et al. (G-9) (bone marrow of the study just discussed as A-8l1, negative control group. Given this portion). Bothinvestigators tested CHA App. 19 is also contained in the record conclusion, I neednot reach more on male Holtzman rats usingfive daily as A-827, App. 13. specific criticisms raised by Abbott. injections. AlthoughLegator used five 6. Deficient Studies. I.have classified (2) Dick, et aI. (A-l77). (a) Study dose levels, Dick matchedhishighest _ the following 15 studies as "deficient" Design: This study was conducted on dose level. Thus, were Dick's findings because they fail to meet the minimum Holtzman rats using CHA as the test credible, theywould indeed place criteria set forth in Subsection C.2. and compound. 14 rats were given daily Lega tor's findings in some.doubt. 3. above. Accordingly, these studies are injections of CHA base at a dose of50 However. as explained above in my . not entitled to any weight. mg/kg body weight for five days. 17 rats discussion of the bone marrow portion I note that all but the first of these were given equivalent doses ofCHA of the Legator study, the Dick studyhas studies (Collin, G-27) were classified by HCL. A negative control group of12 rats a fatal flaw. According to Table ill ofA Abbott as being negative, including the were injected with water. In addition, 177, the incidence ofbreaks and gaps for two studies (Dick, A-l77 and Ford. A two positive control groups were dosed the positive control. TEMis virtually 297J claimed to be exact replicates of for two days. The first group ctio rats identical to that of the negative control Legator's study (G-9). thus, the was injected with triethylenemelamine (water). This means that some unknown weakness of Abbott's position on the (TEM) at a dose of 0.5 mg/kg body factor severely compromised the mutagenicityissue is due in large part to weight, The second group of8 rats experinient's ability to detect breaks the high number of deficient studies on received injections of tris·(2-methly-l and gaps. Thus, since Dick was not even which it relies. aziredinyl) phosphineoxide (METEPA) able to detect an increased incidence of . For organizational purposes. I have at a dose of zo mg/kg body weight. Cells breaks and gaps where theyshouldhave arranged these studies according to the from the bone marrow were analyzed. been, a fortiori no conclusion canbe 61520 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices drawn from Dick's failure to observe study conducted on Holtzman rats. One I find this study to be deficient for breaks and gaps in th~ test compound, group of ten rats received 1%cyclamate several reasons, First, the test CHA. (See general discussion of positive as part of their feed for an unspecified population of only five rats per group Is controls in Subsection C.3.c. above.) period of time. A negative control group too small. Second, no data are presented Due tothls finding, I need not reach consisted of six rats. No positive control (G-124 at 19). Third, no positive control the Bureau's argument regarding the was used. An unspecified number of was used. Therefore, for the reasons insufficiency of Dick's positive control bone marrow cells were analyzed. explained in Subsection C.3. above, I values. (b) Study Results: The investigators attribute no weight at all to this study, '1therefore conclude that this study is reported that the "range of values [found (6) Oser, et al. (A-274). (a) Study entitled to no weight at all because for breaks] is considered to be well Design: This was a multlgeneration Dick's inability to detect a statistically ,within the expected 'background' range feeding study on Wistur rats using CHA significant increase in breaks and gaps of values for normal untreated males of as the test compound. The Fa generation in the positive control, TEM, invalidates this strain and age" (A-195 at 754).No rats were fed doses of 50 or 150 ms/kg any negative findings. data of any consequence was presented, body weight as part of their diet for (3) Ford, et aL (A-297). (a) Study The ALJfound these results to be periods of 6,12 or 18 months. Group , Design: This study was also conducted "negative" (id. at 26). sizes ranged from three to five rats. on Holtzman rats using CHA as the test (c) Analysis: Abbott agrees with the Negative control groups of the same size compound. Three treatment groups with ALJ (Abbott's Exceptions at 56;Abbott's were used for each dose size and period, three animals per group were used. Each Brief at 56).The Bureau makes no but no positive controls. Bone marrow received daily injections for five days. . specific comments on this study. cells were analyzed, averaging about The first group received 50 mg/kg body I find this study to be deficient 250per group. (See Tables 49-50.) weight of CHA base, and the other two because there are virtually no data Cells were also analyzed from the groups were given an equivalent amount which would enable me to evaluate it offspring. For the F1 and F2 generations, of CHA-HCL (obtained from two adequately. Neither is any kind of fetal tissue was taken at Caesarian different suppliers). A negative control statistical analysis presented, nor a section. For the Fageneration, bone group was given water. Two positive' positive control used. (See discussion in marrow was taken from weanHngs. control groups, usirig TEM and Subsections C.2 andC.3 above.) This Negative controls were used for each METEPA,were each dosed for two mutagenicity research was clearly group. but no positive controls. Group days. Cells from the bone marrow were peripheral to the more fully presented and cell populations were as follows: F1, analyzed, numbering 150 per group (250 carcinogenity experiment. Indeed, the 10 rats and 250 cells: Fa,4 rats and 100 for TEM group). authors themselves characterized the cells: Fa.6rats and 300 cells. (See A-274 . (b) Study Results: The investigators at 25a and Table 51). mutagenicity portion as "limited" (A-195 reported their findings as simply (b) Study Results: Cells were scored "negative" for the'CHA treated groups at 752):I therefore conclude that this only for the number and percent of study is entitled to no weight. ' and "positive" for the positive controls abnormalities. The only abnormalities (See Table IV). No statistical analysis (5)Khera, et al. (A-222). (a) Study found were in cells "exhibiting a was mentionedjn"the text or presented Design: The cells analyzed in this study subnormal number of chromosomes" but in table form. . were taken from female Wistar rats these "occurred in no greater proportion As with the Dick study, the ALI found used in a reproduction study. For in the test groups than in the controls" that Dr. Ford "appear[ed] to have used cytogenetic purposes, two groups of five (A-274 at 25c).The investigators the exact protocol used by Dr. Legator rats each were given cyclohexylamine explicitly stated that "[n)o abnormalltles (Ex. No. G-9) but ha[s] failed to . sulfate (CHS) as part of their feed for an in chromatin morphology (e.g., breaks, replicate his results" (id. at 25-26; see extended period ranging from 52 to 67 ~ gaps. exchange figures] were observed" also id. at 35). days. The dose, stated as a percentage in either treated or control groups (A (c) Analysis: Abbott agrees with the of the feed, ranged from 5.56% to 11.12%. 274 at Tables 49-50 arid 51). ALI (Abbott's Exceptions at 72;56: see A negative control was given distilled . The ALI found this study to be also Abbott's Brief at 56-57). The Bureau water. No positive control was used. 100 "negative" (id. at 26). has not made specific comments on this bone barrow cells.from each rat were (c)Analysis: Abbott agrees with the study. analyzed. ALI's finding (Abbott's Exceptions at 50: I find this study to be deficient (b) Study Results: The iilvestigators Abbott's Brief at 56-57). The Bureau because the investigators used only reported "no abnormality in distribution makes no comments on this study. three animals per group. As explained in of chromosome number or incidence of Although this study has some Subsection C.3.a. above, this test is structural aberrations" (A-222 at 267). interesting aspects in its design (i.e., simply too intensitive for any confidence No additional commentary or data was . multi-generation analysis, long duration to be placed in its negative results. presented. of exposure), the study has two fatal Additionally, I find that Dr. Ford did not The ALI found this study to be weaknesses which render it deficient. ' present statistical information '. "negative," but noted that it had been First, with the exception of the F1 demonstrating that the treated groups criticized by the Bureau (id. at 26). generation. all test groups had six or were statistically negative and the (c) Analysis: The Bureau's principal fewer animals. Second, the fact that no positive controls statistically positive. objections were that: (i) the test chromosome abnormalities (such as Along this line, I note that in at least one population was too small; and (ii) no breaks. gaps or exchange figures) were category, "Average percent breaks," the details are given in terms of data, found in any of the groups, treated or incidence for the CHA base group (1.3) background rate, or how the cells were control, raises a serious question about was virtually the same as for the. scored (Bureau's Brief at 85).Abbott the study'e.sensttlvlty, especially positive control, TEM (1.2) (A-297 at admits that "(t)here are reasons for considering the long duration of - Table 1). This study, therefore, is giving less weight to this study," but exposure. This is in contradiction to entitled to no weight at all. ' contends that the investigators did virtually all the other credible studies (4) Friedman, et al. (A-195). (a) Study present adequate data (Abbott's and is a prime example of where Design: This was a cyclamate feeding Exceptions at 58). concurrent positive controls are Federal Register I Vol. 45, No. 181 I Tuesday, September 16. 1900 I Notices 61521 essential. (See general discussion in conclude that no weight at all should be prisoners as test subjects. Sodium Subsection C.3.c.above.] I therefore attributed to it. cyclamate capsules were administered conclude that this study is entitled to no c. Blood cellstudies {humans}. (1) orally for either eight or thirteen weeks. weight at all. Dick, et al. (A-177). For the eight week period, group and b. Blood cellstudies {animals}. 1. (a) Study Design: In this experiment, dose sizes were as follows: 5 subjects, 5 Mostardi, et al. (A-264). four persons (two men and two women) g/day; 5 subjects, 10 glday; 2 subjects, 3 (a) Study Design: This study was were given sodium cyclamate capsules glday (after having 16 glday for 6 days); conducted on Wistar rats using CHA as at a dose of 5 g per day for the men and and 6 subjects, placebo. For the thirteen the test compound. There were two 4 g per day for the women. for a total of week period, these were: 2 subjects, 5 gl treatment groups of three rats each. The four days. These persons had previously day; 3 subjects. 10 g/day; 3 subjects, 3 81 first group received CHA injections at a been tested and found to be able to day (after having 16 glday for 6 days); dose of 20 mg/kg body weight; the convert cyclamate to CHA. In addition. and 4 subjects, placebo. Five of the second group at a dose of 50 mgjkg a similar group of non-converters were subjects were used for both time body weight. Injections were given daily placed on the same dosing regimen. periods. Approximately 10 blood cells for five consecutive days during each of Urine analyses were conducted from each sample were examined. seven weeks. Blood was drawn 24 hours throughout the experiment to verify (b) StudyResults:The authors after the fifth injection of each week. A whether the "converters" and "non reported simply, "No chromosomal negative control group of three animals converters" maintained that status. A abnormalities were observed" (A-703 at was also used. A positive control was negative control group was also final page (unnumbered)). No not. 50 metaphase spreads were established. Blood samples were mutagenicity data was presented. analyzed for each group. obtained and at least 100 metaphascs The ALJfound this study to be (b) StudyResuIts:The investigators (cells) were examined for each sample. "negative" (id at 27). reported "no discernible differences" _The cell slides were coded so that the (c) Analysis:Abbott agrees ....ith the between the treated and controls in person analyzing them did not know ALl's rmding (Abbott's Exceptions at terms of both "the number of abnormal whether they came from a treated or 60-61;Abbott's Briefat 58), emphasizing spreads and percent of cells with control group. that Coulson used dosage levels abnormal chromosomes" (A-264 at 316). (b) Study Results: The investigators comparable or exceeding that of However, no statistical analysis was reported no increased incidence of Bauchinger 0-1) and well above the use presented. Nor did the investigators chromosomal abnormalities in either the level proposed by Abbott in its food explain (beyond the characterization converters or non-converters. What additive petition (Abbott's Brief at 59 "abnormal") how the cells were scored. abnormalities were found were 50).The Bureau criticizes this study on predominantly gaps, with a few breaks. four grounds: (i)no data was presented: The ALJfound this study to be No exchange figures were observed. (il) too few cells (10)were analyzed per "negative" (id at 26). However, one of the "converters" acted subject; (iii) the investigators did not (c) Analysis: Abbott agrees with the as a ''non-eonverter'' during the course specify how the cells were scored: and ALl's finding (Abbott's Exceptions at 59; of the experiment. (iv) the study is unpublished and Abbott's Brief at 57;59).The Bureau The ALJ found this study to be therefore has never been subject to peer criticizes this study on the grounds that: "negative" (ld at 27). review [Bureau's Brief at 90). (i) the test population was too small; [ii] (c)Analysis: Abbott agrees with the I find this study to be deficient too few cells were analyzed: (iii) the ALl's finding (Abbott's Exceptions at because there are insufficient data investigators did not specify what 60-61; Abbott's Brief at 58-59). The presented for evaluation. (See general chromosome aberrations were scored Bureau criticizes this study on three discussion in Subsection C.3.b. above] for: and (iv) there were"enormous" principal grounds: (i) small test In fact, I have reviewed this study in variations in the negative controls from population; (ti) small cumulative dose detail and have found no data at all week to week (Bureau's Brief at 84-85). when compared to Bauchinger 0-1); and relating to mutagenicity. I do note that I find this study to be deficient (iii) inadequate presentation of raw data one expert stated that he thought some because the number of rats per group (3) (Bureau's Brief at 88, relying upon G-124 data were presented (fr. at 526-27). is too small. the presentation of data is at 21;Tr. at 971-72).In its exceptions, That conclusion. however, was based on inadequate in that the investigators do Abbott defends the adequacy of the an admittedly cursory review of the not state for which chromosome data as presented in Tables II and ill of study conducted that same day (id.). A aberrations they scored the cells. and no the study (Abbott's Exceptions at 64). careful review disclosed that the statistical analysis is presented. (See G I find this study to be deficient numerous tables containing blood 124 at 18-19 and general discussion in because of the small population size analyses data related to concentrations Subsections C.2.and 3. above.] I also which consisted of only three subjects of various compounds in the blood (e.g.• note that these findings are not that were demonstrated converters (see protein-bound iodine, thyroxiniodine. confirmed by positive controls. I G-124 at 21 and Tr. at 971-72;see free thyroxin, thyroxin-binding globulin, therefore conclude that no weight at all general discussion in Subsection C.3.a. and plasma cortisol) rather than findings should be attributed to this study. above). I need not reach the issue of of chromosome abnormalities. The (2) Liskerand Cobo (A-241).Although dose size because that would go to the abundance of this irrelevant data the ALJfound that this study "failed to weight of the study had it met the strongly suggests that the chromosome show any positive effects" (id. at 27).it minimum criteria. I also do not reach the analysis was a peripheral part of this appears in the record only in a Spanish issue of the adequacy of the data in study. This may explain why version. This is not an acceptable form Table II of A-177 (Table ill contains chromosome data were not presented. for my evaluation. especially since data on the rat portion of the Moreover, I find the tact that the Abbott has presented no expert experiment). I therefore conclude that no investigators found no chromosome testimony favorably interpreting it. weight at all should be attributed to this abnormalities at all raises a serious Moreover, according to the Bureau's study. question about the study's sensitivity, "Briefat 85. this study employed only two (2) Coulson (A-703). (a) StudyDesign: especially considering the long duration animals (rabbits) per group. I therefore This study was conducted using of exposure. As noted above in 61522 Federal Register I Vol. 45, No.-181 l' Tuesday, September 16, 1980 I Notices connection with the bone marrow note that the cytogenetic portion of this measure a test compound's effectsupon portion of the Oser study (A-274), the study was clearly peripheral to the chromosomes (i.e., breaks, gaps, and absence of any chromosome larger teratology portion, which may exchange figures). The principal abnormalities contradicts the fmdings of account for this shortcoming. Second, no difference between in vitro and in vivo virtually all the credible studies statistical comparison between the studies are that in vitro experiments aru presented in this record and therefore treated and negative controls was performed using cells in test tubes rather presents a second. independentbasis for presented. In fact, there is no indication than live animals (A-BOO at 6: G-124 at classifying the study as deficient. that a negative control was even used. 10). Due to these two major flaws, I need Third, the test population, although The parties agree that the information not reach the other objections raised by unspecified, appears to be grossly derived from in vitro studies is of the Bureau. I conclude that no weight at inadequate. Only ten cells were limited value. The major limitation of in all should be attributed to this study. analyzed per dose level which suggests vitro cytogenetics is that cells in culturo d. Sperm cell studies. (1) Ford {A-297}, that only one or two mice were used per media represent an artificial settIng Friedman (A-195), and Oser (A-274) group. Moreover, this number of cells is which cannot imitate a live unlmal's studies. These three studies have "totally unacceptable" (Tr. at S27).I also metabolism (G-124 at 10 and 30; BOO A already been discussed in connection note that no positive control was used. 800 at 7). Because of this limitation. in with the deficient bone marrow studies, (See general discussion in Subsection vitro studies serve merely as initial Subsection F.a.a. above. In addition, C.2. and 3. above.) I therefore conclude screens to determine if a compound Is each investigator also analyzed sperm that no weight at all should be given to "active" or "inactive" from a cells from male rats. The ALJ found the this study. . mutaggenicity standpoint (G-124 at 25 . sperm cell portions of these studies to (3) LeonardandLinden (A-Z40). (a) and 30; Abbott's Brief at 44). Positive be "negative" (id. at 28). Abbott agrees, Study Design: This was a sodium findings in in vitro cytogenetic studies emphasizing that these and other studies cyclamate feeding study conducted on are therefore Insufflolent, by themselves, rebut Legator's (G-9) positive sperm cell mice. The cyclamate was added to the to declare a compound a mutagen, Such findings (Abbott's Exceptions at 65-£6; drinking water at concentrations of 2.667 findings can, however, buttress more Abbott's Brief at 61). . glliter, S.334glliter, or 10.668 glliter and. definitive in vivo findings, ifpresent. The Bureau offers no additional given to the mice for periods of 30, 60 or That is precisely the situation here, comments. I find that, except for the ISOdays. One mouse was used for each . A review of the in vitro cytogenetJo difference in cells analyzed, the design dose level and time period. Negative studies has shown that the evidence, and reporting of the sperm cell portions controls consisting of one mouse per taken as a whole, strongly suggests that of these studies are identical to that of time period were also established. No cyclamate is "active" from a the bone marrow portions. I therefore positive control was reported. The mutagenicity standpoint. Several studies adopt and incorporate here my previous investigators examined 200 dividing found a statistically significant Increase discussion of these three studies and spermatocytes for each mouse.' of breaks (G-I0,G-l1,G-25(CfU\ conclude that, for the same reasons (b) StudyResults: The investigators portion),G-33,G-35,~9,~6,and stated in Subsection F.a.a. above, each "detected no evidence of chromosome A-722), and one study found such an is deficient and thus should be accorded anomaly. The rate of univalents was increase in exchange figures (G-35). no weight at all. practically the same (±S%) in the Moreover, three of these studies found a (2)Kaziwara, et al. (A-Z17). (a) Study different groups" (A-Z40 at 1-2). No data dose response (G-2S, G-33 and A-722). Design: This study was conducted on relating to the chromosome analysis was These findings outweigh the negative adult male mice using C.H.A. as the test presented. . ones found in the studies relled upon by compound. An unspecified number of The ALJ found this study to be Abbott (A-143; A-205 (calcium mice were in'jected with a single dose of "negative" (id. at 28). cyclamate portion), A-259, and A-300). I CHA, either at 40 mglkg body weight or (c) Analysis: Abbott agrees with the therefore conclude that the in vitro 80 mg/kg body weight. No mention was ALI's finding (Abbott's Exception at 66; cytogenetic studies provide additional made of either a negative or positive Abbott's Brief at 61). The Bureau makes support for my conclusion that Abbott control. Cells analyzed were no specific comments on this study. has not shown that there is a reasonable spermatogonia and primary and I find this study to be deficient for the certainty that cyclamate does not cause secondary spermatocytes. Ten cells following reasons. First, the size of the heritable genetic damage, were analyzed per group. test population (one animal per dose b. The Studies' Findings: Because of (b) Study Results: The investigators level per time period) is totally the limited utility of the in vitro results, I reported only that "[n]o chromosome inadequate. Second, no data relating to will discuss these studies only briefly, aberrations were observed in male the chromosomal analysis is presented (1) Suggestive Studies: The record reproductive cells of mice treated with so/as to allow me to make a proper contains 7 studies which found a either 40 or 80 mg/kg of CHA " (A-217 . evaluation. Finally, I note that: (i) it is statistically significant increase in at 6). No data was presented. No unclear whether a proper statistical chromosome damage which may . statistical analysis was presented. No analysis Was performed; and (ii) no "reasonably be attributed to cyclamate or explanation was given as to how the positive controls were used to validate its metabolites. cells were scored. thefindings. (See general discussion in (a) Stone, et al, (G-I0) found that The ALJ found this study to be Subsection c.z. and 3. above.) I calcium and sodium cyclamate caused a "negative" (id. at 28). therefore conclude that no weight at all statistically significant increase in (c) Analysis: Abbott agrees with the should be given to this study. breaks in human blood cells at ALI's fmding (Abbott's Exceptions at 66; 7. AdditionalSupport: In Vitro concentrations of 2SD-500 meg/mI. Abbott's Brief at 61). The Bureau has not Cytogenetic Studies: a. Summary: The (b) Stoltz, et al. (G-l1) found that commented on this study. parties also submitted in vitro cyclamate, CHA and N-OHCHA each I find this study to be deficient for cytogenetic experiments performed by caused a statistically significant several reasons. First, no data are 13 different investigators. Like the in increase in chromosome aberrations " presented to enable an adequate vivo studies of this class, in vitro (primarily breaks and gaps) in human evaluation of the study. Along this line, I 'cytogenetic experiments are designed to blood cells at concentrations of 10-3, _ Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61523 10- 4 and 10-5 (for cyclamate. equivalent deficient because they do not present Bureau asserts that findings from in to approximately 179. 17.9 and 1.79 meg/ sufficient data for a full evaluation (see vitro studies are relevant only to the ml, respectively; for CHA, 99, 9.9 and Subsection C.3.b. above). Accordingly, question ofwhether the compound is 0.S9 mcg/ml; and for N-DHCHA, 115, they have been eliminated from "active" or "inactive" (Bureau's Brief at 11.5, and 1.15 mcg/mI). consideration. 100: G-124 at ~O). (c) Green, et al. (G-25) found that (c) Analysis: The ALJ made the I agree with the Bureau on this point. CHA caused a statistically significant following conclusion with respect to the As Dr. Legator explained: increase in breaks in rat-kangaroo cells in vitro cytogenetic evidence: I know ofno ....ay in whlch one can with at concentrations of 50, 100 and 500 Most of the in vitro cytogenetic studies. any degree of validitydetermine dosages mcg/mI. A dose response trend was also including the tests on human leukocytes. fromin vitro tests and apply them to in tiro observed (see Table 2). human lymphocytesand kangaroo rat cells studies. When we talk about in vitro testing (d) Kristoffersson (G-33) found that producedsignificantpcsltlve results of we. of course.have a very artificial situation cyclamate caused a statistically serious chromosomll1 aberrations (E'<. Nos. that docs not occur in the animal system, The significant increase in breaks and gaps G-11. G-17, G-ZS. F-33. G-34, G-35. G-39).1n only conclusio:lS that one can make on the in Chinese hamster cells at addition. a statistically significantIncrease in basis ofin vitrostudies is that the compound chromosome breaks and gaps and dose Is active or Inactive,To try to read an;jthing concentrations between 100 and 1,000 dependent results were foundin the studies further into the results. for example. to try to mcg/mI. A dose response trend was also on Chinesehamster cells. Chinesehamster make quantitative extrapolations.is probably observed (G-33 at 278). fibroblasts and humanfibroblasts (EX. Nos. to push the methodfar beyond Uspossible (e) Tokumitsu (G-35) found that G-17. G-33. G-34). Even if the incidenceof usefulness. sodium cyclamate caused a statistically breaks and gaps docs not represent lerioUJ (G-124 at 30). significant increase in breaks and genetic damagea contentionwith whlch I therefore conclude that the exchange figures in human blood cells at many scientists disagree-the presence of a statistically significanteITect cannot be suggestive in vitro cytogenetic studies of a concentration of 0.01M record are relevant and provide strona (approximately 2000 mcg/mI). disregarded. evidence that cyclamate and its 0 (id. at 34: see also id. at 21-:!3). (£) Perez-Requejo (G-37; A-722) found metabolites are "active" in terms of Abbott raises three types of that sodium cyclamate caused a mutagenicity. Accordingly, these studies exceptions to the ALI's findings. First, statistically significant increase in provide additional support for my chromosome aberrations (primarily Abbott challenges the ALI's . conclusion that cyclamate has not been breaks and gaps) in human blood cells characterization of several of the shown to a reasonable certainty not 10 at concentrations of 4.5 and 9.0 mg/mI studies' fmdings [Abbott's Exceptions at cause heritable genetic damage. (equal to 4500 and 9000 mcg/mI). A dose 4~5 and 68-69). As is evident from my response was also found (A-722 at 5). description of these studies' Iindlngs, I G. OtherStudies Insufjicient to (g)Ebenezer (G-39) found that sodium agree with Abbott that only Tokumitsu Out....ei;gh Suggesli\'e Bvidence cyclamate caused a statistically (G-35) found a statistically significant 1. Summary: In addition to the in \7'.'0 significant increase in chromosome increase in exchange figures. Moreover, cytogenetic studies discussed above. the aberrations (breaks. gaps and fragments, my finding that the Dixon study (G-34) record contains three othertypes of in grouped together) in human blood cells is deficient and that the Meisner study viva mutagenicity studies: (a) host at concentrations of .02 and .04 mg/mI , (A-259] is negative dismisses any mediated assay: (b) dominant lethal (equal to 20 and40 mcg/mI). concerns that Abbott may have with the assay; and (c) drosophila. Several (2)Negative Studies: The record also ALI's characterization of those results. I additional vitro tests were also contains four studies which found no disagree, however, with Abbott's in performed. The studies from each of statistically significant increase in attempt to dismiss the results of the chromosome damage which may Stoltz study (G-11) (increased these groups produced predominantly reasonably be attributed to cyclamate or incidences of breaks) due to negative results. These firttlings. its metabolites. "cytotoxicity." Cytotoxicity means cell however, are insufficient to outweigh fa) Brewen, et al. (A-143) found no death. As explained above, it is true that the strongly suggestive in \7\'0 CHA or N-OHCHA induced increase in chromosomes with breaks will cytogenetic findings discussed above because known mutagens have been chromosome aberrations in human sometimes di~ rather than repair blood cells at concentrations of 20, 100 themselves or join with other broken found 10 show mutagenic effects in some and 500mcg/mI CHA or 25, 50. 100. 200 chromosomes to form exchange figures in vivo test methods but not in others and 250mcg/ml N-OHCHA. (see Subsection F.2.a. above). However, (G-124 at 9-10 and 31; Tr. at 933-34; Tr. (b) Green, et al. (G-25) found no findings of breaks are nevertheless at 498-501; Tr. at 717-18 and 734: SEe calcium cyclamate induced increase in biologically significant for the reasons discussion in Subsection D.1. above), chromosome breaks in rat-kangaroo set forth in detail in Subsection F.2.t. and because the last group of in vitro cells at concentrations up to 200 meg/ above. Abbott's exception that this studies are by their very nature ml, study is insignificant due to observed insufficient to outwelgh suggestive in (c) Shamberger. et at (A-300) found cytotoxicity is therefore without merit. vivo findings (see Subsection A.3 no significant increase in breaks in Second, Abbott asserts more directly above), human blood cells treated with sodium that breaks do not constitute serious 2. Host-MediatedAssay: The ALJ cyclamate treated in concentrations of genetic damage (Abbott's Exceptions at made the following findings with regard 100 meM (approximately 20 mcg/mI). 68-69). Again, I have already addressed to the four host-mediated assay studies: (d) Meisner, et at (A-259), found no this issue extensively In subsection The host-mediatedassa;,' is a mutagenicity statistically significant increase in F.2.c. above and need not repeat it here. test whlch involvesplacing a kn!)~\..n breaks in human fibroblasts after Finally, Abbott claims that the indicator organisminto the interperitoneal exposure to cyclamate in a positive findings were achieved only cavity of a treated animal, considered a host. The host animalis then treated v.ith the test concentration of 500 meg/mI. through the use of massive doses which compound,in this case:sodiumor calcium (3) DeficientStudies: Three studies, are not relevant 10 human experience cyclamate or CHAo Upon conclusionof the Schoeller et al. (G-18); Dixon (G-34) and (Abbott's Exceptions at 46; Abbott's testing.the indicator is removed and Lederer, et al, (G-46; A-235) are Brief at 51: A-800 at 8). In response, the examinedfor mutations.The primary 61524 Federal Register I Vol. 45, No. 181 I Tuesday, September 16. 1980 I Notices advantage of the host-mediated assay is that has terr implanted embryos, three of , (ii) L~rke (A-827, App, 15) found no it provides the sensitivity of in vitro tests which later died, the subsequent statistically signficant increase in pro- or combined with exposure to a metabolic examination of the uterus will reveal post-implantation loss after mating . process as in in vivo tests. Mice were used as seven live embryos and a total of three NMRI/BOM strain male mice (dosed the host animal and either Salmonella dead embryos or rasorptions, The with a total 50 g/kg of cyclamate over 5 typhimiriumor Serratia morescens was used as an indicator. Chinese hamsters were used mutagenic significance of post days) with untreated females. in one experiment usinghuman leukocytes as implantation loss is that it is caused by (iii) Lorke, et al, (A-827. App, 9) found an Indicator. The results of testing both chromosome damage, such as exchange no statistically significant increase in cyclamate and CHA were negative in the four figures (G-29 at 128;A-151 at 472). pre- or post-implantation loss after . studies conducted (Ex. Nos. A-143. A-26B. A A second, less significant factor to be mating NMRI male mice (dosed with a 325, A-375). looked for is called "pre-implantation total of 750 mg/kg of CHS over 5 days) (Id. at 24). Abbott takes no substantive loss." This means that an embryo has with untreated females. exceptions to this portion of the-ALI's died before it has implanted itself in the (iv) Lorke, et a1.(A-827. App, 17) opinion. Abbott does note, however, uterus. This figure is obtained by found no statistically significant that all the sentences but the final one subtracting the number of implant sites increase in pre- or post-implantation ,"appear without cites." Abbott therefore in the uterus (both viable and non loss after mating NMRI strain male and suggests that these constitute "non viable) from the number of "corpora female mice. Both sexes were treated for substantive statements" rather than lute a" in the ovaries (i.e., sites from ten weeks prior to mating. Doses were "finding[s]" (Abbott's Exceptions at 50). where eggs were shed) (A-827, App. 9 at either 2,000mg/kg/day of sodium The Bureau makes no comments on 9; A-827, App. 15 at 4). Pre-implantation cyclamate (1% of feed) or 200mg/kg/ Joss is less significant from a day of CHA (0.11% of feed). these studies. (v) Lorke, et al, (A-827. App, 11) found I adopt the above-quoted statement of mutagenicity standpoint because. given the state of scientific knowledge. it is no statistically significant increase in . the ALJ and agree that these four studies. pre- or post-implantation loss after are negative. I also find that the ALI's not certain that such losses are necessarily due to genetic damage (A mating NMRI strain treated female mice description of the host-mediated assay 827, App. 15 at 8; G-121 at 14J. (single dose of 10 g of sodium method is amply supported by the texts The dominant lethal assay, in one cyclamate) with untreated males. The of the studies themselves, and that respect, is an excellent mutagenicity test· published version ofthls study (A-011. therefore the ALI's citations are method because it enables one to App. 18) shows that a positive control adequate.Thus, these statements examine the mutagenic effects of a test (cyclophosphamide) was used and that constitute substantive findings. As noted compound on progeny (G-124 at 11). In positive dominant lethal results were above, however, negative studies using another respect, however, this method is obtained. the host-mediated assay are insufficient quite limited because it only measures (vi) Ford. et al. (A-297) found no to outweigh the suggestive cytogenetic "lethal" effects; thus. non-lethal statistically significant increase in post experiments (see discussion in mutagenic effects. which may still be implantation loss after mating male Cox Subsection Da. above). serious. go undetected (id at 11-12). Swiss albino mice (given a single 3. Dominant LethalAssay. a. b. The Studies'Findings: Findings injection of 50/mg CHA) with untreated Description ofTest Methods: A from the is dominant lethal studies are females. These negative findings were dominant lethal assay is a study described below. One of these studies confirmed by statistfcally significant designed to detect geneticall caused produced findings suggestive of positive findings in several positive deaths in the nexf(F1) generation. The mutagenicity (G-29), and nine studies control groups (see Table II). ' study is conducted in three principal produced negative findings. Also (vii) Epstein. et al. (A-182) found no steps: (1) dosing the animals; (2) described briefly below are the five statistically significant (P<.05) increase allowing the animals to mate; and (3) implantationloss was found (see Table" several grounds: (1) small number of that these results were not replicated by 2).The only positive findings (p<.05) in animals: (2) a t}'Pically low number of other investigators using other strains CHA-treated animals was in terms of implanted and live embryoes in and species is a separate issue to be pre-implantation loss, but the authors untreated controls; (3)fmdings not discussed below. concluded that this result was not of replicated by any other researcher; and Abbott's final criticism results from a genetic origin (A-2oo at 33). (4) the ALJwrongly said that the misinterpretation of a statement made (ti) Kennedy, et al. (A-220) (rat dominant lethality observed by Peterson by the ALJwith respect to this study. portion) found no statistically significant increased over time (Abbott's The statement in question is as follows: increase in post-implantation loss after Exceptions at 52-53). Abbott's fU'St two However, the effect seen in the CHA mating Charles River albino male rats points are based on a brief evaluation of treated animals was significantly higher than (dosed with 1.5 or 15.0mg/kg day of this study by Larke, et al. contained in that of the saline control which increased CHS for 60 days) with females of the one of their dominant lethal studies o...er the thre.e or six weeks {sic]period. same strain (similarly dosed, but only described above (A-827, App. 9 at 10). (Id. at 25). Abbott suggests that this for the 14 days immediately prior to The Bureau defends the Peterson study statementwrongly implies thatPeterson mating). by stating that: (1) a small animal f3)DeficientStudies: For the following found an increase in dominant lethality population is adequate where the over time. I agree with Abbott that reasons, five studies which reported findings are positive: (2) the raw number negative findings are deficient and Peterson did not fmd sucha time-related of live and dead implants are less increase. However, I do not interpret therefore entitled to no weight. important than the ratio between the Friedman. et al. fA-19S) fed male that ALI's statement to convey this fact. two; and (3) Peterson was able 10 Rather. I interpret the ALl's statement to Holtzman rats a diet of2%calcium replicate his own results, even ifother cyclamate and then mated them with mean that Peterson found an increase researchers were not (Bureau's Brlef at over both the three and six week untreated females. The presentation of 81-82). data in this study. however. is periods, not that the findings in the sixth With respect to the.small number of week were greater than those inthe inadequate because there is no animals used, I agree with the Bureau comparison shown between the number third week. This is consistent with the that where a study's findings are facls and should satisfy Abbott's of dead implants per female and the positive, a small animal population does total number of implants per females exception. not negate the validity of the study (see I therefore conclude that the Peterson (see A-195 at 754).Khera, etal, (A-221] Subsection e.3.a. above). However, I also did not provide adequate data to ~tudy is strongly suggestive of also agree with Abbott to the extent that mutagenicity, especially sincehis support their finding that CHS did not a small animal population detracts cause a statistically significant increase findings were replicated. I would have somewhat from the weight to be given to more confidence in these results, in post-implantation loss. The only data that study (id.). I therefore consider the presented arein a rough graph (Figure B] however. ifPelerson had used more Peterson study to be facin1lyvalid but which lacks the necessary precision to animals, if the total number of implants entitled to slightly less weight than permit an adequate evaluation. Finally. had been greater, and ifhis findings had the two Oser studies {A-273and A-274) would similar results from a larger been replicated by an independent and the rabbit portion of the Kennedy animal population. investigator. study (A:-220) were actually teratology The second issue regarding the total (2) The Larke Studies {A-827. Apps. g, studies rather than dominant lethal number of implants being atypically 11. 15 and17}. Abbott places great experiments. A teratology study is small is also a question ofweight rather reliance upon four studies conducted by where the females are treated with the . than validity. As Dr. Green explained, Larke, et al, As described above. Larke test compound during pregnancyto the threshold issue in a dominant lethal used several different procedures, determine ifany effect is produced on study is the ratio between the li"'ing and including not only the traditional the growing fetus. Because dosing takes dead implants rather than their total method of mating treatedmales l\ith place after conception, this type of study number (G-12.3 at 5).The Bureau does untreated females. but also the less can not possibly detect mutagenic admit, however, that the total number of common modes of mating untreated effects on germ cells prior to conception implants per female was low (Bureau's males with trealed females and of (as is the purpose of a dominant lethal Brief at 81): thishas the effect of treating both sexes before mating. All study). The fact that these are indeed reducing the "test population" (id. at 81. four studies produced negative findings teratology studies is reflected in the n. 20)for purposes of statistical analysis, for post-implantationloss. descriptions of test methods (see A-273 and hence reduces the study's The Bureau attacks the validity of at 9-10, A-274 at 6-7 and A-220 at 6-7). sensitivity. Indeed, as Dr. Legator. these studies on two grounds: (1) that Thus, these studies are not entitled to stated: It. •• one of the serious Lorke failed to perform preliminary any weight in the evaluation of the shortcomings in the Peterson study was experiments necessary to determine the potential dominant lethality of the low number of implants per female" "maximum tolerated dose" to be used in cyclamate or its metabolites. (Tr. at 948-49).Thus, again. I consider the dominantlethal studies on c. Analysis: The studies in dispute are this study to be facially valid but cyclamate (Bureau's Brief at 93-94) but Peterson, et al, {G-29),the four studies entitled to somewhat less weight than rather used only mathematical by Lorke, et al. (A-827. App. 15: A-827, would similar results from a study with extrapolations (Bureau's Reply at 26); App. 9; A-827, App.17; and A-827, App. more implants per female. and (2) the alleged failure to use positive 11), and Epstein, et al, (CHA portion) The third issue regarding replicability controls in the experiment where the (A-18Z). requires only brief discussion. I agree females were treated rather than the (1) The Peterson Study (G-29). This with the Bureau that Peterson did males (Bureau's Brief at 94).Abbott suggestive study, as noted above, was replicate his results with similar defends these studies by asserting that the only dominant lethal study to statistically significant (p<.05) findings Larke did properly ascertain the observe statistically significant (p<.OS) of post-implantation loss (G-29 at maximum tolerated dose (Abbott's Brief positive findings of post-implantation Tables II and III). This replicability adds at 69; Abbott's Exceptions at 54) and loss. Abbott criticizes thisstudy on credence to Peterson's findings. The fact that a positive control was used in the t> 61526 Federal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices study using treated females. (Abbott's (3) The Epstein Study (CHA portion]. (G-123 at 5). I therefore conclude that, Brief at 70]. A review of the record The parties agree that the findings of although most of the dominant lethal shows that Abbott is correct on both . post-implantation loss in this study are' evidence is negative, some question still points. . not statistically significant at the P.;;;.05 remains about the mutagenic potential The,maximum tolerated dose ("MTD"] level. Based upon the testimony of Dr, of cyclamate and CHA in at least one is "the dose just below [the one in] Epstein, however, the Bureau maintains strain of mouse. Thus, the evidence is which one sees obvious toxicity" (Tr. at that borderline findings of pre not conclusive. However, even were the' 847].The parties agree that in a implantation loss make the study evidence conclusively negative in tho dominant lethal study it is important to "suggestive" rather than "negative" (Tr. • dominant lethal studies, such findings use the MTD in order to maximize the at 865-66]. I disagree. As noted above in would be insufficient to outweigh the chances of detecting a positive effect. the description of dominant.lethal assay . suggestive cytogenetic experiments [sea The parties interpret differently, test methods, findings of pre Subsection Da. above). Indeed, the however, the following statement in one implantation loss, even if statistically dominant lethal assay technique has of the Lorke studies which describes significant at the P.;;;.051evel, are not been known to report negative findings how the dose level used was necessarily tied to mutagenicity. Even for compounds that arc proven ascertained: Dr. Epstein admits this (G-121at 14; Tr. mutagens in other test methods (G-124 This dose was chosenbecause preliminary at 866), as does Dr. Green in his at 9-10; Tr. at 498-500). experiments had demonstratedthat it is we1l dominant lethal study which did find a 4. Drosophila. The final type of in vivo tolerated by the animals.The administration statistically significant (P.;;;.05) increase mutagencity testing performed on of higherdoses wouldhave created in pre-implantation loss (A-206 at 29).~ cyclamate or its metabolites was considerabledifficulties due to the large conducted using Drosophila (fruit Illes), ~ubstance Indeed, in Dr. Green's study, he quantityof involved. concluded that the.pre-implantation loss· The specific type of Drosophila test (A-827, App. 8 at 4]. This statement was not of genetic origin (A-206 at 33). which was conducted is called a "sex makes clear, first of all, that Larke used This doesnot mean that findings of pre linked recessive lethal" test. The ALI "preliminary studies" and not . implantation loss would never be described this test as follows: "Mathematical extrapolations" to considered biologically significant, but A recessivelethal mutationpresent on a ascertain the properdose. Second, I corroborating evidence would be male's onlyX chromosome wlll cause the interpret Lorke's statement, when read needed (such as statistically significant ( male to die.If the compond beingtested as a whole, to mean that the findings of post-implantation loss in the inducesa recessivelethal, and the affected "maximum" dose arrived at was . same study). Accordingly, I have . gene is carried In the X chromosome of the maximum in terms of potential toxicity. I attributed no weight to the fmdings in sperm. the matingwith untreated femoles will therefore reject the Bureau's criticism produceoffspring (FI), whichwhen mated this record of pre-implantation loss. togetherproducemales (F2),half of which -that Lorke did not properly determine (d] The Evidence As a Whole. The the MTD. Moreover, I note that 'the total have X chromosomes fromthe original ALI found that the positive findings in treated males.If thisgroupIs absent lin the dose used by Lorke in the CHA study the Peterson study (G-29) are not F2generation] recessivelethals Were using treated males (A-827, App. 9) completely rebutted by the negative produced. {approximately 510 mg/kg) 26 was mouse studies because of the difference (ld. at 23; see G-122 at 8-10 for more comparable to that used by Cattanach, in mouse strain tested: et al. (A-151] (500mg/kg). I therefore detailed description). The sex-linked find that Lorke's dose levels were In evaluatingthe results ofvarious tests. it recessive lethal test is "the most adequate. . . mustbe remembered that various strains efficient and informative procedure in / react with various degreesofsensitivityto Drosophila testing" (G-122 at 8), It will The second issue regarding postltive chemical mutagens, • controls may be disposed of easily. The detect a wide range of genetic damage, Bureau complains that one specific (ID at 25j. The ALJ therefore concluded principally in the gene mutation study (A-827, App. 11] in which that "the results cannot be disregarded" category (id. at 11). untreated males were mated with (lD at 35]. Abbott takes exception to this The Drosophila evidence in this treated females lacked a necessary finding of the ALI and maintains that the record consists of two negative studies positive control. Although it is true that dominant lethal studies, when viewed in (A-712 and A-728) and five deficient no positive control information is the aggregate, arenegative (Abbott's ones (G-24, G-122 at 20, A-263, A-209 reported in the unpublished version of Exceptions at 71; 53). and A-305). The two negative studies this study (A-827' App. l1),1he . As is evident from the above require some discussion because tha published version (A-811, App. 18] description of the dominant lethal Bureau has questioned how much shows that a positive control studies, most of the evidence in this test weight should be attributed to them. (cyclophosphamide] was used and that method are negative, and these studies Vogel, et al. (A-728) and Knapp, at lit. positive results were obtained. (A encompass several mouse and rat (A-712) each conducted sex-linked comparison of the data in Table 1 of A strains. Nevertheless, none of these recessive lethal-tests as described 827, App, 11 with that in Table II of A negative studies used the same strain of above. Vogel, et al, conducted adult 811, App, "l8 shows that they are indeed (C57Bl/Fe] mice as did Peterson. The feeding tests using sodium cyclamate the same study.) record is clear that differences in strains and CHA as the test compounds. Knapp, I therefore conclude that the four are important. As Dr. Green explained: et a1.conducted adult injection and studies by Lorke are all negative in . larvae feeding tests using the terms of post-implantation loss and are Therefore. when one considersthe fact that metabolites CHA and N·OHCHA as the the strain ofmouseutilizedby Petersonwas test compounds. The ALI found both of .entitled to considerable weight. not employedby the other investigators, one has to constderthe possibilitythat the effect these studies to be negative (ld. at 23), "This study actually employed CHS rather than observedby Petersonet al. Was genuine. This and I agree. CHA. The total dose of CHS was'750 ms/kg. The The Bureau maintains, however, that authors stated. however. that 150 mg of CHS equals study raises the possibilitythat approximately 102 mg of CHA base [A-827. App. 9 cyclohexylamine can producedominant . neither of these studies has a large at 4). The total dose.when converted to CHA. is lethalityin animalspossessingcertain enough test population to establish therefore approximately 510mg/kg. geneticconstitutions. safety for this test system (Bureau's Federal Register I Vol. 45. No. 181 I Tuesday. September 16, 1980 I Notices 61527 Brief at 91-92; Bureau's Reply at 23). The Moon. et al, (A-263). Browning (A-Zoa) involving negative Ames test Bureau based this position on the (discussed in G-122 at 20). and Roller, et results on calcuim cyclamate. CHA and testimony of Dr. Zimmering, who al. (A-289). All of these studies .....ere N-DHCGA (Abbott's Exceptions at 40 explained that a population size of available only as abstracts without the 41). Similarly. Abbott c«;!rrectIy observes 12,000 X chromosomes (F1 flies) in both data necessary for a full evaluation (G that in the second paragraph discussing the treated and control groups would be 122 at 21; see Tr. at 484). Moreover, I Ute Chu study (G-47), the ALI failed to necessary to achieve a test sensitivity note that for the two abstracts which expressly state that the CRA portion of capable of detecting a doubling over the reported positive Iindings (A-3OSand the study was negative (id. at 41) control rate (G-122 at 13; Tr. at 485-86). G-24). Commissioner Kennedy asked in However, Abbott incorrectly suggests Dr. Zimmering considered this degree of his Remand Order that theparties that the ALJ omitted to cite negative sensitivity to be necessary to establish supply more information (44 FR 47623). findings from studies A-736 and A-808 safety because Vogel. et al. found the The parties have since stipulated that (id. at 42). for citations ofthese studies frequency of recessive lethals in the the requested data is unavailable are contained in the AL]'s first treated group ofBrood 3 (0.68%) to be (Stipulation dated September 17, 1979 at paragraph quoted above. I find these roughly double the frequency of 5-7). I therefore am attributing no weight minor omissions by the ALI to be recessive lethals in the controls (0.36%) to these five studies. inconsequential. (G-122 at13). Although this difference In summary, the available Drosophila Finally, Abbott contends that the AL} was not statistically significant for evidence is negative, but the sensitivity was wrong in one instance. With respect Vogel's population size (approximately of these studies is such that they do not to the Chu study (G-47) (second 17{)(h3200 for all Broods combined). a establish the safety of cyclamate and its paragraph), the ALI stated that the cell test with a population of 12,000 would metabolites in this test system. Even survival rate was reduced after detect positive findings ifthe relative were those studies to establish safety in ''prolonged treatment," while Abbott frequencies between the treated and this test system. however. evidence in contends that the cell survival rate was control groups remained the same (id.). Drosophila would be insufficient to reduced at ''increased concentrations" Thus. Dr. Zimmering would require the outweigh the cytogenetic findings (see (Abbott's Exceptions at 42). A review of larger experiment to test his hypothesis Subsection D.l. above). G-47 shows that Abbott is correct on (id. at 13-14). He noted that Drosophila 5.AdditionalIn Vitro Testing. The this point. All this means. however, is tests of this size are "carried out final category of mutagenicity evidence that the increase in mutations was seen routinely in most laboratories" (ld. at contained in the record involves at increased concentrations rather than 13). Dr. Zimmering make a similar additional in vitro testing performed on after prolonged treatment. The study, analysis with respect to the Knapp study cyclamate or its metabolites. As noted therefore, still reports positive (id at 16). above. however, in vitrostudies by their findings.» I agree with the Bureau on this point, very nature are useful only as In summary. these in vitrostudies were predominantly negative, although but I emphasize that the issue goes to preliminary screens and cannot two investigators did find positive the issue goes to the weight to be outweigh positive or suggestive in vivo results in studies that were not directly attributed to these studies. not their findings (see SubsectionA.3 above). rebutted. ThUs. the evidence is not validity. The studies as carried out and The ALI made the following findings conclusive. Even were these studies reported are valid negative studies. Dr. with respect to these in vitro test: conclusively negative, however. such Zimmering's point, with which I agree. is In Vitro Tests. The Ames test has would be insufficient to simply that given the frequencies of rmding previouslybeen described. Addltional results outweigh the suggestive using the Ames test were introduced in vitro recessive lethals found in these experiments (see SubsecUonA.3 above). experiments. much larger tests would be concerningthe mutagenicityIssue.Both necessary to establish safety in this test cyclamate and CHA were tested by several H. Miscellaneous Mutagenicity Issues system. scientists usingSalmonella typhImuriwn. All the results werenegative (Ex. Nos.A-738,A 1. TheRelationshipBetween The Bureau also contends that the 008. G-124). However. positive results were Mutagenicityand Cancer. The ALI Drosophila evidence is incomplete in found using CHAin Saccharomyces cerevisla found that "[m]utagens in somatic cells that no experiment tested cyclamate (as (Ex. No. A-268). can lead to cancer" (ld at 35). thereby opposed to the metabolites) using the In addition, Chinesehamster cells were suggesting a causalIink between adult injection method [Bureau's Brief at cultured with CHA or N mutagenicity and carcinogenicity (see 92). Again. the Bureau relies upon the hydroxychlohexylamine [N-QH-CHA)added also Id, at 21). The Bureau agrees testimony of Dr. Zimmering (G-122 at in a study to examine gene mutation.A (Bureau's Brief at 71-72 and Bureau's forward mutation change was seen with N 14-15). Although Dr. Zimmering's Reply at 21). Abbott, however. takes testimony is quite persuasive as to why OH-eHA (Ex. No.G-47). Uponprolonged treatment with N-QH-eHA the cell survlval strong exception to this finding by the each different route of administration rate was reduced to ~ and a significant ALI both as a matter of general must be used. he does not explain why increase in mutations was seen over the scientific principle and as applied to the the parent compound (i.e., cyclamate) controls (Ex. No.G-47). evidence on cyclamate (Abbott's must be tested using each such route Plant cell studles were also performed Exceptions at 39-40 and 74). where. as here. the metabolites have using onion seeds or HawortbJa (Ex. Nos. A A review ofthe record in this already been so tested. and where 250. A-251.A-295). The results of the plant proceeding shows that adquate expert cyclamate itselfhas been tested in an cell studles on sodlumcyclamate were testimony was not elicited as to any of adult feeding study. I therefore reject negative. the issues concerning the relationship this criticism raised by the Bureau. (ld at 21). Abbott's exceptions to this between mutagenicity and The record also contains five portion of the ALI's opinion primarily careinogenicity-e.g.•what, ifany, types Drosophila studies which I have found involve clarifications rather than to be deficient, all due to an inadequate disagreements. For example. Abbott • As notedIn Subsection F.2.c. (2) above. presentation of data (see Subsection correctly notes that in the first however. since thisstudy Is reported ani)"as abstract. both ils posllive fmdings with N-OHCHA C.3.b. above). These studies are Stith. et paragraph discussing the Ames test, the and neptive findings with CHAare entitled10 al, (A-305). Majundar, et al, (G-24). ALI failed to cite fidings by Dr. Legator Uttle.1f any.wd8ht. 61528 Federal Register I Vol. 45,'No. 181 I Tuesday, September 16. 1980 I Notices - of genetic damage cause cancer; what, if mooted by the conclusions I have does not contain (1) sections entitled any, types of mutagenicity study results already reached with respect to "findings of fact" and "conclusions of would serve as an indicator that the test carcinogenicity and mutagenicity. law;" (2) a full articulation of the compound may cause cancer; and The two issues of acceptable daily reasons for the findings and conclusions finally, the applicability of these issues, intake and safe conditions for use are that are made, and (3) full citations to if any, to the evidence of mutagenicity. interrelated. The acceptable daily intake the record (Abbott's Exceptions at 3-0). Given the inadequacies of the record in level is the level (expressed in mg/kg A careful review of these exceptions this respect, I make no fmdings body weight/day) immediately below leads me to conclude that they ga concerning what, if any, relationship the lowest level which produces primarily to form rather than substanco, exists between mutagenicity and cancer. significant adverse or toxic effects. For I therefore find that there is no merit in ~. Findings ofthe Temporary cyclamate, the parties introduced the argument that the Initial Decision Committee. The Temporary Committee, evidence concerning testicular atrophy does not comply with 21 CPR12.120(b). in its Review of Data on the and reproductive effects. Once the It is not necessary for the Initial Carcinogenicity of Cyclamate, make acceptable daily intake level is Decielonto contain a detailed several findings with respect to the determined, the probable consumption discussion of every item of evidence in mutagenicity studies (G-41 at 32-36)., patterns of cyclamate must be order to have evaluated it adequately; These findings provide additional calculated to determine whether, If nor is it necessary to provide a record support for my conclusion that the cyclamate is added to the food supply as citation for every factual statement in eVidenc~ in ~i~ record, particularly the Abbott proposes, actual consumption the decision so long as the decision is cytogentic stu;dies, stro~ly suggests that # would exceed the acceptable daily supported by the record. Although the cyclamate or Its metabolites may cause intake level. This latter calculation is the main, text of the Initial Decision is brief heritable genetic damage. Indeed, the safe conditions for use issue. For the in its explanation of the reasons for Temporarr Committee concluded a~. purposes of this discussion, it is not resolution ofthe scientific issues, I find follows With respect to the mutagenicity necessary to state precisely how these that the Initial Decision's discussion of studies: calculations are made. the issues and citations to the record urn .•. the fact that severallaboratorieshave The ALI found that the administrative sufficient both to support the ultimate shown that cyclamateand cyclohexylamine record would support a finding "that the findings and conclusions made, and to can producechromosome d~mll:ge ~ both acceptable daily intake is five mg adequately inform Abbott of the reasons rodent,s an~ humansfolIowmg ~n Vl~O r cyclamate/kg bodyweight/day or less" for those findings and conclusions. admlmslrahon,?fdoses appr.o~ating (Id at 38J.This is consistent with the Moreover, it is clear from the ALl's humanusageraises the possibility that these ,.. lth ugh bb detailed description of the studies compounds may adverselyaffectgenetic' Bureau s posttion, a 0 A ott activity. advocates a higher level. On the second submitted that the ALI examined tho ~e foun~ record in detail. Accordingly, the Initial (Id t 36'\ 30 issue, ALI that since each . a 'J party either overestimated or Decision complies with 12.120[b). VI. Acceptable Daily Intake and eafe underestimated the probable B. AllegedFailure To Comply With 21 Conditions for Use consumption figures to support its U.S.C.348 c~ Two additional issues were addressed respective po.sition, "neither be Abbott contends that the Initial by the parties during the hearing phase relied on to give an acc.ur~~e picture of cons~ption Decision is not a "fair evaluation of the of this proceeding. The ALI described the probable of cyclamate record" in that the ALI unfairly these issues as follows: (ld. at 37). Accor~~ly, the ALI,found that the safe conditions for use Issue evaluated the evidence. Thus, Abbott [1.] Apa~t ~m the [carcinogenicity and was not resolvable on this record (Id at asserts that the ALI failed to comply mutagenicity] Issues..., what does the 38) with section 409 of the act, 21 U.S.C. 348 evidentiaryrecordshowis an acceptable '. I th th .. I' (Abbott's Exceptions at 3). daily intake revelfon:yclamate? It IS C ear at e questions mvo vmg an~ I find that this exception is also [2.]Whetherapart fromthe acceI!t.able daily intake safe . without merit. In most respects, this [carcinogenicity and mutagenicity] issues conditions for use are only Important if •.., because ofprobableconsumption Abbott prevails on both the exception faults the Initial Decision simply because the decision did not patte~s, safe conditions of use can be carcinogenicity and mutagenicity issues. prescribed. This is because under the act, as accept the arguments offered by Abbott (e.g., finding a study "suggestive" even (ld. at 4). As explained in more detail explained in Section II. above, the though the results of the study are not below, I find it is unnecessary to decide agency must deny approval of Abbott's significant at the .05 level). Abbott's either of these two issues since they are food additive petition if Abbott fails to specific arguments concerning the ALI's , prove either that cyclamate is not unfair evaluation of the evldeneuare :IllThe only significant finding made bythe carcinogenic or that cyclamate is not Temporary Committee that isatvariance with my mutagenic. Since Abbott has failed to discussed in detail in the body of this mutagenicity findings relates tothe dominant lethal make either of these two showings, I decision. I fmd that the AL} did carefully .assay evidence. The Temporary Committee reported consider Abbott's arguments. See, e.g., that "there isnoevidence thateither cyclamate or find it unnecessary to decide the ID at 5, 6, 7, 8, 9, 13, 14,15, 16, 17, 19, 22. cyclohexylamine possess dominant lethal effects" . acceptable daily intake and safe (G-41 at33).1. however. found that one study by conditions for use issues. 25,26-27,28,31,32,.36, and 37. With Peterson, et al, (G-29) contains statistically . minor exceptions discussed in the body significant (P=.05) fmdings ofpost-implantation VII. Miscellaneous Matters of this decision, the ALI correctly 10SD (see Bubsectlons G.3.b.(1)(a) andG.3.c.(1) evaluated the evidence. Although the nbove).llis quite possible. however. that tha A. Allegations Conoeming 21 CFR Temporary Committee never reviewed this study 12.120(b)· Initial Decision does contain some since the Temporary Committee docs notspecify errors, virtually all of these are any dominant lethal study byauthor. andsince the Abbott made several general inconsequential. They clearly do not Temporary Committee reviewed only 11 dominant objections which relate primarily to the reflect any prejudice or unfairness in lethal studies whereas the hearing record contains form of the Initial Decision. Abbott 15'.1 therefore conclude that this fmding bythe evaluating the evidence, but rather an Temporary Committee docs notnecessarily contends that the Initial Decision fails to impartial and conscientious effort to contradict the findings made inthis decision. comply with 21 CFR12.120(b)in that it resolve the issues. I Fedeeal Register I Vol. 45, No. 181 I Tuesday, September 16, 1980 I Notices 61529 C.Allegations That the Initial Decision and recommendations which clearly do oplnlona. madea technicaljudgment be was Is a Repudiation ofScience not fall within the category of "factual authorizedto make. Abbott further contends that the information:' As the preamble to I therefore reject all ofAbbott's Initial Decision is a repudiation of Subpart B of FDA's Administrative exceptions concerning FDA's internal science in that it reflects a lack of Practice and Procedure Regulations deliberative process (Abbott's Remand understanding of the scientific evidence explains: Ex. at 4-11). and rejects fundamental principles such •••The Commissioner advises that the E. Separation ofFunctions as statistical significance, repllcabillty, requirement of this section [21 em12.85] presence of uncontrolled variables. and does not extend to documents reflecting the FDA regulations governing the scientific peer review (Abbott's agency'sintemal deliberativeprooesl. e.g.. conduct of agency officials in Exceptions at 6-12). Almost all of these documents ex:presalng the point ofview of administrative hearings, such as the general exceptions are discussed by agencyemployees who reviewedan NDA. cyclamate proceeding, provide that: even though such documents are containedin Abbott in connection with Abbott's • • • Repretentativesof thebureau shall and admlnlstrative filerelatingto a matter not participateor advise in any decision criticism of the ALI's evaluation of that is the subject of the hearing. specific studies. I have therefore except as witne..or counselinpublic discussed those significant exceptions in (41FR 51714.November 23.1976). proceedings. Thereis to be no other communication betweenrepresentativesof detail in connection with my evaluation Indeed, section 12.85(a)(2) has since the bureau and representativesof the of each specific study. In general. been amended to express clearly this Commissioner concerning the maUer Abbott's contention that the Initial longstanding agency interpretation: (involved in the hearing] beforethe decision Decision is a repudiation of science is "Internal memoranda reflecting the of the Commissioner. without merit. Although there were deliberative process * • * are not 21 CFR :10.55{b}(2}(i). Abbott complains some minor errors in the Initial Decision required to be submitted" (44 FR 22344, that this regulation was violated and in some instances its phrasing could April 13. 1979). because Dr. Vasillios Frankos and Dr. be improved. when evaluated on an I therefore find that the Bureau did not Constantine Zervos, who presently work overall basis. the limitations of the act improperly in withholding the in the Commissioner's Office ofHealth Initial Decision do not undercut the documents at issue. In any event, the Affairs. have served as advisers to the validity of its basic finding that court in Abbott Laboratories v. Harris, Bureau of Foods in this proceeding. Dr. cyclamate has not been shown to be Civil No. 79-C-3732 (N.D.m, decided Frankos also served as a witness for the safe. That finding is supported by a July 12,1980) flatly rejected Abbott's Bureau. Abbott contends that Dr. large body of scientific studies and claim that these documents show that Frankos and Dr.Zervos might "taint" expert testimony contained in the its food additive petition would have other scientists who are responsible for record. been approved had not then advising me (Abbott's Remand Briefat D. Documents Relating to the Internal Commissioner Schmidt included 25-26). Abbott further contends that Dr. Deliberative Process improper considerations in making his Zervos and a Bureau attorney may have decision: contacted other scientists in the Office Abbott moves to admit into evidence of Health Affairs and "tainted" them by two sets of documents which reflect the The picturewhichmergesfromthe record is one ofgoodfaith uncertaintycaused b.1' the asking them to advise the Bureau (id.). decisionmaking process that led to the limitations ofprior testingand differing !he requirement ofseparation of agency's decision in 1976 to deny interpretatlonsof the results.Manyin the functions is designed to ensure that the approval of the food additive petition for scienlificcommunity believedand believe samepersons do not serve as both cyclamate. For purposes of limiteduse of cyclamatesto be safe to a advocate and judge in the same identification. the first set of documents reasonable certainty;others have not been proceeding. Thus, in the context of the is attached to Abbott's Exceptions to the able to so conclude. Virtuallyno one Is of the cyclamate hearing, representatives from Initial Remand Decision, dated February opinionthat the limitedute of cyclamatesII demonstrably unsafe. the bureau of Foods (the "advocate") 25, 1980;the second set is attached to a are forbidden from having certian Motion to Include Documents, dated Giventhole circumstances. it i. not surprising that therewere differences of communications with representatives April 17, 1980.Both sets were obtained opinionamongadvisors and that, in collegial from the office of the Commissioner (the by Abbott through civil discovery discussion, views changed.The initialdenial "judge"). This restriction is intended to ordered in Abbott Laboratories v, was viewed.as stated by one witness.as a "avoid even the appearance of Harris, 481 F. Supp. 74 (N.D.m, 1979). very closecall. a very difficult judgment. unfairness" (40 FR 40691; September 3, Abbott argues that the Bureau should Possiblytheviews ofsomead\ilors were 1975).At the same time, the restriction have disclosed these documents to influencedby theirperceptionof the on communications is limited to "the Abbott in accordance with 21 CFR Commission"'. tentative judgment. Possibly matter" which is involved in the hearing. 12.85(a)(2). the Commissioner. despitehis recognition of the properlegalstandard, was himself Here, that matter is the substantive I have reviewed these documents in issue of whether cyclamate has been their entirety and find that they do not somewhatinfluenced by his own perception. of the need or lack ofneed. forcyclamatesin shown to be safe. fall within the purview of 21 CFR the marketplace. Withoutdoubt publicfocus The mere fact that two former 12.85(a)(2).That section provides that, upon and plainUrrsinterest in the question representatives of or advisors to the prior to the issuance of a notice of caused the decisionmaking process to be Bureau (Dr. Frankos and Dr. Zervos) hearing, the director of the responsible somewhatmorecautiousand ponderousthan now workin the office ofthe bureau shall disclose: it otherwisemighthave been. Commissioner is insufficient by itself to All documents in the director'sfiles The recorddoes not,however.supportthe constitute a violation ofseparation of containing factualinformation. whether conclusion that defendanll and their functions. The regulations prohibit favorableor unfavorableto the director's predecessorsacted in bad faith or for improperreasons.Rather, the depositionof certain communiations, not mere position, whichrelate to the issues involved proximity of offices. Abbott has not in the hearing.••• the Commissioner makingthe initial decision reveals a somewhatacerbicgentlemenwith presented any credible evidence that The documents at issue. however, strongviews and a willingness to express either of these two scientists has had contain internal, pre decisional opinions themwho.after considering numerous substantive communications regarding 61530 Federal Register I Vol. 45, No. 181 I Tuesday, September 16,1980 I Notices the' safety of cyclamate with any person determining whether cyclamate has advising me on this issue. . . been shown to be safe. Neither has Abbott demonstrated that VIll. Conclusion either Dr. Zervos or a Bureau attorney contacted other scientists on my staff to Based on the foregoing findings, ask them to advise the Bureau. Even if conclusions, and discussion, I affirm the such communications were made, ' Initial Decisions and conclude that: however, I do not consider them to 1. Section 409(cJ(3)(A) of the act, 21 violate separation of functions because U.S.C. 348(c)(3)(A);reqJ}ires FDA to the communications would not have deny approval of.a food additive involved any substantive discussion on petition ifa fair evaluation of the data the safety of cyclamate. presented fails to establish that the food Finally, Abbott complains that the additive will be safe under its proposed separation of functions regulation use. See Section n. . prohibits Bureau attorneys'from 2. "Safe" means.a reasonable representing the Commissioner in a certainty of no harm. See Section n. lawsuit filed by Abbott which sought a 3. The act places the burden of .declaratory judgment that Abbott's food proving safety on the company seeking additive petition be approved (Abbott's approval of the food additive petition. Remand Brief at 27). That lawsuit has See Section n. not involved an evaluation of the 4. For Abbott to obtain approval of its evidence on cyclamate's safety, but ' food additive petition, it must prove that rather allegations by Abbott concerning the data in the record establish that whether improper considerations played there is a reasonable certainty of no a role in the agency's prior decision to harm from the proposed use of deny approval of Abbott's petition. cyclamate. See Section n. Since separation of functions does not. 5. The data in the record do not apply to the latter subject, it is establish that there is a reasonable appropriate for Bureau of Food's certainty that cyclamate does not cause attorneys to participate in that lawsuit. cancer. See Sections ill and IV. I therefore reject all of Abbott's 6. The data in the record also do not contentions regarding separation of establish that there is a reasonable functions. certainty that cyclamate does not cause heritable genetic damage. See Section V. F.Admissibilityofthe IRLG Report 7. Abbott has failed to meet its burden The ALJ refused to admit into of proving that cyclamate is safe under evidence exhibit G-142, the report of the its proposed use. See Sections ill, IV and V. ' Interagency Regulatory Liaison Grou~ ("IRLG"), on the ground that the 8.lD light of these findings and document was not in final form and was conclusions, the Issues involving thus subject to further alteration by the . acceptable daily intake and safe FDA (IRD at 5). As the ALJ explained, conditions for use need not be decided. the report contains a discussion of See Section VI. scientific concepts and methods The foregoing decision in its entirety concerning the evaluation of substances constitute's my findings offact and that may pose a risk of cancer in conclusions of law. . humans (id.). The document in its IX. Order current form is a government proposal subject to public notice and comment In accordance with subsections procedures (id.). The Bureau takes (c)(3)(A), (£)(1) and (£)(2) of section 409 exception to this ruling, contending that of the act (21 U.S.C. 348(c)(3)(A), (£)(1) lack of finality is not a proper basis for and (£)(2)) and 21 CFR 12.130, and under the exclusion of evidence (Bureau's the authority delegated to the Remand Ex. at 5-6). Abbott urges me to Commissioner (21 CFR 5.1J, the food uphold this ruling (Abbott's Remand additive petition (pAP 4A 2975)for Reply at 7). approval of cyclamate for use as a I agree with the Bureau that the . sweetening agent in food and for . report's lack of finality goes to weight technological purposes in food is denied. rather than admissibility. The IRLG The Initial Decisions are affirmed, as report is therefore admitted into modified and supplemented herein. evidence because its purported subject In accordance with section 409(£)(3) of the matter is relevant. However, because act (21 U.S.C. 348(£)(3)), the effective date of the document contains preliminary this order is December 15. 1980. views only, and because, in any event, Dated: September 4,1980. the Bureau has not adequately shown Jere E. Goyan, exactly how these views should be CommissionerofFoodandDrugs. applied to this record, the document has [FRDoc.80-27590 Filed 9-4-SO; 1:54pm) not been given any weight in BILUNG CODE 411D-03-M