Tobacco Documents Online

-5- curves that is shown in the graphs. It will be noticed that for the 180 DMBA, and for 6 of the 7 ti~e points for the 120 IMBA group, the confidence reKion of In (turnout rat£o) is not plotted. These confidence limits were not plotted because they could not be estimated, and this siCuatlon arose because the slope of the dose/response curve did noc reach siKnlficance. In Tables IB to 2IB the slope of. the condensate dose/response line is tested for sisnificance using a *'rule of thumb" t-test. The ratio of the slope to ice 8tmadard error is treated am s t-statistic with (N- 3) deKrees of freedom, where N is the number of tumour-bearins animals used in tstlmatinS the slope. Tables 1B to 21B show that only for the 60 DMBA Kroup at week 36 onuards, at reek 48 for the 120 DMBA Stoup, is this t-statistic siKnific~nt at the 95Z level; for the 180 DMNA Kroup it never attains siKn~ficance. The results are shown plotted in FiKures 4 to 10. Where po£ncs are noc shown (for example, 60 DMBA, 20 mK in Figures 4, 5, 6) the number of cumour-bearln8 =~-ls in these groups at the alms-point of the plot is zero. Thus it can be seen that it is not until the analysis at week &8 chat there are turnout-bearing animals present in all the 24 Kroups. Tables 1B Co 21B also show Chat, for the 60 DMBA Stoups zha difference between the condensates B9/I and B916 is siKnif£cJnt from week ~A onwards (actual sisnlficamce level at week 44 - 94.60., 95:[), for t:bm 120 Kroups the betveen-condene~tte effect is always s~nlflcant, and for cht 180 ~ Kroups the coudemsace effect 4s s:LIp~ficamt fz'om week AO ouvards. Thus alr3,oujh ~ slope of the condemsate close-:espouse curve £s, except for the 60 DMBA S:otrpz, lll-de~tamd a difference between the condensaCes is readily detectable. FiEures 4 co 10 quite clearly BAT Co LTD - MINNESOTA TOBACCO LITIGATION k..rl ................................................................................ --'-'r-

INDEX HAS INDICATED GAP IN BATES RANGE HERE

-6- illustrate the situation. Rice often in these fi|ures the dose-response line £or the 120 and 180 D]4KA Kroups is near horizontal or hag a ve~-y shallow slope. On the other hand the seneral ~evel o£ response of the animals to the two ¢ondensates £s always clearly di£Serent, Bg/1 pz~ducL~K am obviously hiKher response than B9/6. DZSCUSSIC~ Two questions are o£ interest:- (a) Is there an optimum level of Dm~ pre-t-reatmant? (b) CouZd the experlmsnt have been teru~ated belore reek $7? (a) FiSures I. 2 and 3 and the analyses of variance, shov quite clearly that, 8enerally sp~tklns, only the 60 Dt4~A 8~oups provide a satlsfactoz7 ulysis. Only at week &8 do the ~20 ~ ~roups provide an estimate of the tumor£Se~ic ratio, and the 180 INiKA 8~oups never do. As noted previously, this ar£ees because the condensate dose-response curves of the 120 DNBA and 180 DKBA cannot be estabZ~shed. This seems to me to ~nd£cete that these latter two lemtls o5 ~ preotreetmsnt are too hlgh. The e~£ect o5 the ~ in4t:Letor has overud3alJned the e~£ect o5 the condensate promotor to such an extent that all condensate dose levels look al~e. All that remains to be seen 4e the ve~-y larks d£55erence betveen the condensatas. The B9 lank term sir'in paiztt~g experiment is now complete, and an initial ana1?sls of unvalidated data gives a tumorisenic ratio 5or )9/1 to B9/6 o5 5.31, i.e. the dose level o5 B9/6 must be about 3.31 times the dose ~evel of B9/]. to produce equal response Ln the animals. The results o£ this interim mmlys£s o£ the B9 long-term experiment are also shown plottnd on lt4ruz~ns I to 3, and inspection o5 these ££Kures shows that £or all the ~ 8Temps the esti~tte of the t~smoz~gen4c ratio is fairly comparable with the lon4;-tetm BAT Co LTD - MINNESOTA TOBACCO LITIGATION ........................................................................ ---.v

INDEX HAS INDICATED GAP IN BATES RANGE HERE

-7- result, but only for the 60 DI~A groups does the slgpe of the condensate dose/response curve become comparable ~th that of the lon4~cerm test, and then only from about week ~ ouvards. Thus the conclusion to be drawn from this analysis seems to be that the 120 ~g and 180 ~g levels of D~ pre-tremCmeut ee certaLuly too h~sh. but whether or noc the 60 uZ level is optima.l is difficult Co establish. Tf it is intended to repeat an experiment of Chlm nature, then • wuld suggest thmc DHBA l~cer-treatmmnt levels of 30 VS, 60 ~K, and 90 US be tried. (b) Obviously. as far as the 120 DMBA and 180 DMBA ~roups are concerned very little would have been lost by carac~c~nK the ozper~nmnc earlier. In general for these groups the condensate dose/response curve is never established and the confidence region for the Cumocigenic ratio cannot be estimated. For the 60 ~ IFcoups however the situation is far from clear. Although a confidence rqion for the tumor/~m~c ratio could usually be calculated, the width of the region varied mnormously, this variacien being governed by the pattern of incidence of the Cumour-bearinl animals. At weak 44 a good analysis was obtained, but st the two succmedin8 t~mm poLucs (weeks 48 and 52) the upper l~mit of In (c~tour ratio) bec~e very high (30.05 and 18.74 respectively). This sharp ~acraaae tea the e upper limit m occasioned by the fact that the numbers of t~mour-baaring 8n~mals occurring with the a9/l condensate continued to increase wlth ttme, whereas the number of tumour-bear/n~ animals occurr/ng with the B9/6 condensate bec~me static from weeks 44 to 52. &c week 57 the incidence of turnout-bearing animals increased in the B9/6 E~oups and the upper l~It of the In (Cutout ratio) was correspondingly reduced. BAT Co LTD - MINNESOTA TOBACCO LITIGATION J ¢...r'l %.0 ............................................................................... -T-

INDEX HAS INDICATED GAP IN BATES RANGE HERE

-8- T~ practice 0£ course it £s the lover 14m~C of_ln (CUnour rttlo) vh£ch is of £nceresc, whe~ In CCtmour reclo) is sruCer Ckan zero i.e. vhen the cumorisen£c ratio is K=eacer chart 1.0. Zf the ~over I~n~c of In (Surtout ratio) is SreaCer Chart sere Chert the t~mor£ge~ic rac£o £s 8izn~££cantly Kremcer Chsn 1.0. ThLs s£cuac£on occurs for the 60 DHBA groups after week 44. Thus aC any time on. or afcer week 44 the experia~nc could have been term/hated and the 60 DMBA SToups would have shown the tumorigenic =atio between the coudensmtes to have been significant • Much of the fluctuations in the Krsphs shown in FLsures I co 3, e.g. the exceptionally high upper limit for In (Cumour retie) at weeks 48 and 5 in ¥£Sure 1, is due to the fact Chat there ~tre only 21 m~dmAls per group £. e. at etch level of condensate dose and ~ pre-t~eatnnmsC only 21 animals vere used and thus the 4ncldence o£ only ]L or 2 hey tumour-besr£uK enLmals has • marked effect on the m~ly8£8. ~ larger poups of m~m-ls had bee used then the inc4dmnce of ctunour--bear~aq~ ~le is usually smoother and the situation as ac weeks ~8 and 52 in Fisure 1 would have been less l£kely Co ar£se, so thac the effect of ~s:ly termLnatiou on the 60 DHJA ~oupe (or the other ~HBA levels) would have been clearer. The overall conclusion seems to be therefore t~st there -.=,y be some advanCase in terms of t~ne saved (and consequently ~ Co:ms of cost of experiment) Lf the 60 DHBA Stoups had been cermtz~ted on or after the 44Oh week. But 4f £t is intended Co adopc th~s procedure £n future exper/mencscion the nmnbers of animals used oushc co be considerably increased so as Co scab£14se the incidence o£ cumour-bear£uZ BAT Co LTD - MINNESOTA TOBACCO LITIGATION O

INDEX HAS INDICATED GAP IN BATES RANGE HERE

-9- anLmals. As a rule of thumb Z would sussesC us£n4 e~ least 80 anLmals per Stoup, and scopp£ng the exper~aenc when about IOZ o£ the animals have become cumour-bearins An the Eroup showing r.he lowest response. assumes thac morcallty effects rill be relatively 8mmll, CONCLUSIONS 1. ~e o ~e Thls Dose levels of 120 ~S and 180 ~g of DMBA as prrtrut~:ts in an in~tia~or-promotor experiment are too h~Eh. A dose level of 60 ~S of DH~k produces useful data, but 4t yes not possibie ¢o esr~tblieh that this m the optimum dose level. Xarly tetz~Lnation of the experiment (i.e. before week 57) may be advantaseous if the 60 ~S DHBA level of /nltiator is used. The determ~uatlon of the stopping time would be made easier by the use of larKer ~:oups of animals. ..KEFFJLEHCES 1. "The dtsip and analysis of an experiment to compare rJ~e ~mour- proaott~K activitles o£ the condensate from ciKarettes B9/1 and B9/6". E.B. Wilkes. G.l. S D.C. Keport No. RD.11$2-u, October 1974. 2. "Pseudo analysis of variance o£ mouse skin ptintinS emperimmnts uslnS • ge£bull dlstrlbutlon with known k and v". P.N. Lee. Z.R.C. Docmaent E828, RlO03, 1970. 3. ''We£bull dlstr~but:ions £or continuous catc/JBolenesls ezperCmonts". R. PeSo and P.H. Lee. Biometrics 29, 3, pp. 457-470. BAT Co LTD - MINNESOTA TOBACCO LITIGATION .-,.,j

INDEX HAS INDICATED GAP IN BATES RANGE HERE