Tobacco Documents Online

& COM~A~KSON OF THE TUMORIGENTC ACTIVITIES OF JANUS CONDENSAIES BO, B2 AND BA EEPOKT NO. RD.1537 RESTRICTED 24.10.1977 AUTHOR: E.B. Wilkes ZSSUED BY: N.E. Willis PROG. ~. : D ZSTRIBUTTON: Dr. S.J. Green Copy No. I Dr. D.G. FalCon " " 2, 3 ~ibrar7 " " 4, 5 COPY NO.: mmma~ CO mam~ BAT Co LTD - MINNESOTA TOBACCO LITIGATION

~-sWtC,~/461>-Z Group Eeee•rch & Development Centre, British-American Tobacco Co. Ltd., S0~N. 24ch October 1977 A COMPARISON OF THE TUMORIGENIC ACTZVITIES OF JANUS CONDENSATES BO, B2~ AND BA. (Report No. RD.1537 Restricted) SLaY This report describes a re-analysis o£ the early JANUS txpe~nts. The &nalysis o£ the ¢•l£br&C£on groups seems to suggest that there was e siKni£icanc loss of sensitivlty o£ the experimental animals during the course of experiments BO to BS. This apparent effect could however be due to • loss of act£vity of the standard carcinogens during storage. This latter conclusion v•s supported by the results from Ohm J~NUS repeat experiment B5 wh£ch showed no £all in the eati~stas o£ the tumoriKeni¢ actlvlCy o£ the condeusetes which were re--exmLLned. On this basle results £or condensate obtained from a cased and £1avoured U.S. blend (B2) are compared with those from a £1ue-cured all-lam£na bland (BO) and • lamina-stma blend (34). Some of the data from the N.C.I. series Ill experiment has •leo been rra~alymed. The results of this re-analysl8, combined with the results £rom the JANUS ezperimenc8, lead to the concLusLon that there ks no raison to belleve thac the incorporation of cocoa into the blend of a c~sareCce leads to an Lucruse in the cumorilmn£c activity of the smoke condensate. ~t is also concluded that the use of susar and/or humeccant has no ef£ecc on tumorigenic accivitT. I mmm=h BAT Co LTD - MINNESOTA TOBACCO LITIGATION

-2- I. ENTRODUCTION This comparison has been made in an attempt to assess more fully the implications arising from some of the conclusions drawn from the analysis of the Series ZIZ N.C.Z. 8kln palnting experiment (I). 2. THE DESIGN OF THE EXPERIMENTS The comparison of these condensaCee is hampered by the fact that they were not directly compared in 8 81nEle experiment, nor was a control condensate used to fom a link between successive experiments in the JANUS series. The relationship between these condensatee must be determined by =onsiderinE the results obtained from the calibration groups, and the results from experiment B5 wherein condensates BO and B2 were directly compared, but only at a single dose level. ~n the early JANUS experiments (BO to B4) calibration groups of animals were used. In each experiment there were & calibration groups treated as follows: The first calibration group was painted three times per week ~th O. 1 mZ of palncin8 solution contelninE 20 ~K of 3,4-benzpyrene. This painting continued throughout the first 13 weeks of che experiment and then ceased. 2. The second group was similarly treated except that the painting ~onc£nued throughout the first 26 weeks o~ the experiment. 3. For the thLrd calibration group the O. 1 ml of paint4n8 solution solution conta/~ed 20 ~g of 1,2, S,6-dibmnzanthracenm and the palntinS was continued Chroushout the first 13 weeks of the experiment. h. The fourth calibration group was slmilmrly crested to the third except chat the paintinE continued throushout the first 26 weeks of the experiment. m r~ Go r~D o BAT Co LTD - MINNESOTA TOBACCO LITIGATION

-3- Iu Kouer81, each of these Kroups contained 27 animals, but any an/J~al dying durinK the £irst ~ weeks of the experLmont va8 replaced whenever possible, 8o that the number of animals entering the experiJnent is occasionally KreaCer than 27. THE RESULTS FROM CALZBRATXON CROUPS BO TO B4 The data are eho~ Ln Tables 1 to 20. Experiments BA and BS shared a common batch of animals and so only one set of ¢allbrat£on Stoups was used for those two experiments. Zn Tables 1 co 20, T is the experlme~c t£me £n weeks, N £8 the number of annuals as risk oc she beginn£ng o£ the time period, D£ £s the number of animals respo~d£ns during the time period (£.e. the number o£ animals becom£nK t,m~ur-baaring, or the number o£ animals becoming malignant cumou~-bear£nK) and D2 Is the number of animals dyin8 vithouc respondins. Since the animals used in these calibration Kroup8 ~ere not painted th~ouKhout their Lifetime the simple Weibull model i8 noC appropriate for the enelys£8 of th48 data. A6e stsndardismt£0u procedures were therefore used So obca£n t~nour rases corrected for mor~l~cy end these aKe-scandardised turnout rates are shown in Table 21. The ase-scandardlaatlon procedures used (Lee actuarlal method) £8 described in (2). The 8~andard aKe-speclfic mortallcy rates used are shown in Table 22. The aSe-standardlsed turnout rates of these callbrat£on Stoups are 8hou~ plotted 4n Y£Suro8 1 to 4 and from chase jraph8 £t 48 clear that not onl7 £s the response to the two carcinoKens different, and that the duret£on o£ treatment has on of£oct on response, but also that there ha8 been a £a11-o££ in the response of the exper£mmntm2 an~T.ma18 to those standard carcinosens. The analys£e of var£ance shown in Table8 23 w w BAT Co LTD - MINNESOTA TOBACCO LITIGATION

-4- and 74 confirm this visual assessment of the data. The between carcinogen effect and the between duration effect ere both highly sisnifican~, and there is no siEnlficant carcinogen x duration interaction. The ~inear time trend is hiEhly siK~/ficant, and for the all turnout data there is some evidence of a norr-llne&r component in thR time trend. The coefficients for Time (linear) in the model used were, for ell tumours -.05781, end for meliEnant tumours -.04271. For the purposes of fittinK the s~del the starting times for Bl, B2, B3, and BA were measu=ed £n days after the start time for BO. Since the arc sin square root transformation was .sea (to homoEenise the variances) these coefficients are in units of sin-I 4response rate per day, the response race beinK expressed as • fraction of unity rather than as a percentaEe. The fitted lines are also drawn on FiEures I to 4. This analysis of the calibration Eroups indicates that for the early JANUS experiments, successive batches of animals showed a diminishing response to the standard carcinogens benzpyrene and dibeu- anthracene applied to the shaved dorsal skin of the animals. For the experiments studied (BO to B4) this fall-off in response 18 l£~,aar with t~na, after the arc Sin square root varienc@ s~tb£1~s£nK transformation has been used. THE KESULTS FROM CONDENSATE TRY.ATED GROUPS BO TO B5 The data are shown in Tables 25 to 39, and the eKe--sr~sndmrdf.sed ttmour rates ere shown in Table 40. (JANUS experiment B1 is omitted since the puff volumes were varied durinK condensate production. The BI condensates are therefore not directly comparable with those of other experiments. ) rag.rob CO r~ BAT Co LTD - MINNESOTA TOBACCO LITIGATION

-5- Figures 5 and 6 show plots of the aKe-standmrdised turnout rates of the reported experiments. These were" (a) BO, dose level 50 ms, repeaeed in BS. (b) B2, dose level 50 ms, repeated in BS. (c) B3, dose level 50 ms, repeated in BS. From these figures it £s apparent that there has been no faLl-off in the response of the animals towards the smoke conden~ates. Thus we are left with ~ contradictory observetlons. On the one hand the calibration Kroups show 8 clear 1o8s of response of the 8signals towards the standard carclnosens benzpyrene and dlbenzanthracene, but on the ocher hand the response to the condensates BO, B2 and B3 has not chnged. In (3) the fall in response of the calibraClon groups was noted, and it was concluded that the activity of the standard carcinogens had fallen i.e. under the conditions of storage being used, the dibmzzan~hracmnm and the benzpyrene (kept both as solid and £n solution) was deterlorandns. Ocher, more complex, explanations of PiKures I to 6 are of course possible, but £n the absence of supporting evldlnce for an alternative explanation the m/~ple hypothesis of loss o£ mcCivIL-y of the scmnddrd chem~cmls seems the most rmasouab~Le to adopt. Figures 7 and 8 show the &Ke-scandardlsed turnout races of the four coudeusaces SO, B2, B3 and B4 plotted against dose level. From chase graphs it can he seen thac there ere d£fferences becwemn the condansaces, and Chac the highest dose level used (75 ms three t£mss per reek) was probably Coo high in Chat the "high dose anomaly" is displayed by all four curves. That is Co say, the response eC the high~sC dose leve~ £s less than ~uld be expected from :he responses observed at the two lower dose ~LevelLs. mmmlp r~J BAT Co LTD - MINNESOTA TOBACCO LITIGATION

The differences between the =ondensares were 8x~n~ned using the Weibull distribution function. Using the methods described in (4), c~n values of k and w, and separace values o£ b for each group of animals, were E£tted to the data o£ Tables 25 to 39. Estates Og the parameters were obtained for all tumours and for malignant tumours only. The est~zates of b, k and w, vlth the associated 8~umry stat£stLcs S and V, are shown in Tables A1 end 42. The goodness-of-fit of these parameters to the t~nour 4ncldence data was tested via the usual X2 s~atistic and the results are shown in Tables A3 and 4A. The values of X2 obtaLned (9.630 for all t~mours, 7.1A3 for m=lignant turnouts, both with 6 degrees o£ freedom) are noc significant at the 95X level of eonf£dence, indicating ti~l: a satLsfaeto~-y ££t has been achieved. Tables 4S and A6 show the results of an analysis of var£enee derived £roa the l£kel£hoods of var£ous We£bu11 models fitted to the data. This technique £s gully described in (A). These a~alyses sho~ that the dif£erences bet~nan the coudensates are hilhl7 s£S~Lficant, that the dose levels used have a h£Khly s~Knificant e£fect, and that there is no siEn~ficant condensate x dose interact£on. The partitlon£ng of the dose ef£ect shows ~hat bo~h the dose linear and dose non-l£near terms are highly sLgn£flcant. The tumorigen£e ratios show that all the pair-~rise di££eranees bet~an the four condansatas (BO, B2, B3, BA) era significant, and that the ranlo8 derLved from the all-t~our amalya£s are 4n good agreement with those obta£ned from tho analTsis of maLiKr~nt turnouts. The tumor£sen~e ratios are Lnterpreted as £ollovs. For ~le, the rat£o for BO v. B2 (all turnouts) is 0.771. This means that a dose m BAT Co LTD --- MINNESOTA TOBA.CCQLITIC;ATION

-7- imvel of iOO mg of condensate B2 would produce the same t~r£Ke~c response in a group of animals as a dose o£ 77.1 ~ of condensate BO (other dose levels pro race). This interpretation assumes thac the dose-response relationship is ILnear over the range of doses cons Ldmred. Figure 9 however indicates thac there is some evidence of a high dose anomaly. That is to may, the response of the animals at the highest dose ~evel is somewhat less than expected a£tmr due allowance h;Ls been made for the e££ects o£ morcallcy. The analysis of variance shown in Tables A5 and A6 congirm the presence o£ thls anomaly in Che sense that the dose non--llnear term is highly s4~flcamt. This non- linearity o£ response will lead to a bias £n the estimate o£ the slope o£ the dose-response line when the linear model is £itted to the data tar the purposes o~ computLn~ the tumox~Kenic ratios. Thus a better ear,mate o£ these ratios may be obtained by omitting the hiah dose level data from the analysis. The tumo:igenic =atios obtained by omitCin8 the 75 ms dose level dace are shown in Table 47. A couwsr~son o£ Table 47 with Tables 45 and 46 shows that the e££act o£ omitting the h~gh dose level data Ls to reduce the di££erences between the condensates i.e. the tumor4Kam£c ratios have all become closer to 1.0. As a result, one of the comperiao~s, B2 v. BA tar malignant turnouts, is no longer e£Sni£icanc at the 95Z level o£ ccm~4dence. When considering di££erences between these condenaates, the ratios og Table 47 are to be pra£et~red to those o£ Tables 45 and A6. BAT Co LTD - MINNESOTA TOBACCO LITIGATION ........................................... ]

-8- 5. THE RESULTS FROM THE N.C.I. THIRD SERIES OF CIGARETTES The data, taken from (I), are shown in Tables 48 co 57. It will be noted from Tables 48 and 53 that groups 3 and 4, and Stoups 31 and 32, were all painted an Uhe low dose level (12.5 mE). In the analysis that follows groups 3 and 4 were pooled to form e sinE'le group, as were groups 31 and 32. Similarly, groups 7, 29, 51 and 53 are all repeats of the control blend (SEB ~II with casing) at the low dose level and 8o were pooled to form a single group, and the group 8, 30, 52 and 54 were pooled to give • single high dose level (25 mE) group for the control blend. A X2 nest based upon likellhoods derived from Welbull parameters shows ChaC this pooling is in general justified. There was some indication chac the four low dose level groups for the control blend were differing in response, group 51 being suspiciously low. (These four groups were nonetheless combined to form a single group.) The age-standardised t~ur rates are shown in Table 58. The actuarial method due to Lee was used, and in order to minim/me the corrections due Co the different mortelit£es of the various groups the age-speciflc mortality races were derived from the mean mor~allty of the groups being compazed. These age-specific mortality rates are shown in Table 59. The data of Table 58 are shown plotted in Figure IO. The conclusions drawn by N.C.I. are summarised 4. Table 60 (Oaken from (1)), m~eompanied by Ohm followLng c~ents= "SEB ITI rich powdered cocoa ranks 14tb out of ~he 19 va~£ables eesced st the I2.5 ms dose Level and 19tb of the 23 variables tested at the 25 ms dose level. These low rankln6s suggest than powdered cocoa contributes co the tumorigenicity of the condensate. t,,u r~ CXD BAT Co LTD -_MINNESO_ TA__TOBACCO LITIGATION

-9- SEB lZZ with no sugar and SEB ZI~I with no hmnecCmnt bol:h have about the same ranking as SEB $II at each dose level, ms&seating thst sugar by itself or humectant by itself has little or no effect on condensate tumorigenicity. On the ocher hand, SEB ~TI" ~Ch neither sugar nor h,wnectant has a ranking comparable to SEB ZIZ at the 12.3 n~ level, but ranks let among all 23 variables tasted at t~e 25 mS level. This suggests thac at the lower dome, sugar and humectanc in combination have little effect on condensate tumorigenicity but, at the higher dose level, sugar and humectaut in combi~acioa may contribute to condensate tumoriKenicity." Eeferr£ng to 7iKu~e ~0," these couclusions ms7 be summa=ised as: B is more active chat C, E, Y or G. C, E, ¥, and G are the same at the low dose level, but Y ks less active than C, E, or G at the high dome level. (~n passing, it must be noted that the 19 variables tasted at the low dose levml, and the 23 tested at the high dome level, refmrced to by H.C.Z., arm the full set of variables ezmninad in their experiment. The data in Tables 48 to 57 are a subset of the data given 4n (I). This subset forms the basis o£ the N.C.~. coucZusious rosardin8 the tumorigenic activity of cocoa, sugar, and humectant. ) The N.C.I. conclusions glvm much weight to the low response of F at the 25 mS doe level, and interpret Chls effect as some kind of synergistic efface of 8user and h~mectant. I cannot accept these conclusions because (a) £t is difficult to i~agine a cumorigouic effect of sugar and hummctant much that it is demonstrable at the 25 q dose level but not at the 12.5 mS Level. mmm.~ ~D C0 BAT Co LTD - MINN~TA TOBACCO LITIGATION